Immunity, Vol. 7, 715±725, November, 1997, Copyright 1997 by Cell Press
Early Lethality, Functional NF-?B Activation,
and Increased Sensitivity to TNF-Induced
Cell Death in TRAF2-Deficient Mice
both TNFR complexes. Whereas TRAF2 interacts di-
rectly with TNFR2 (Rothe et al., 1994), it is recruited to
TNFR1 via its interaction with an adaptor protein called
TRADD (TNFR1-associated death domain protein; Hsu
et al., 1995, 1996a). TRAF2 also interacts with CD30
(Ansieauetal., 1996;Gedrichetal., 1996;Leeetal., 1996;
Boucheret al., 1997)and CD40 (Rothe et al., 1995a), two
additional members of the TNFR superfamily involved
in immune regulation (Smith et al., 1994).
TRAF2 is the prototypical member of a family of six
intracellular signal transducers characterized by a con-
served C-terminal TRAF domain (Hu et al., 1994; Rothe
et al., 1994; Cheng et al., 1995; Mosialos et al., 1995;
Regnier etal., 1995; Rothe et al., 1995a; Cao etal., 1996;
Ishida etal., 1996a, 1996b; Nakano et al., 1996;Takeuchi
et al., 1996). The TRAF domain of the TRAF2 protein
is further subdivided into a highly conserved TRAF-C
domain that is required for TRAF2 recruitment to the
receptorand a coiled-coilTRAF-N domainthat interacts
with two cellular inhibitors of apoptosis, c-IAP1 and
c-IAP2 (Rothe et al., 1995b). With the exception of
TRAF1, the N-terminalregion of each TRAF protein con-
tains a RING finger structure and multiple zinc fingers,
domains thought to dictate the signals generated by a
particular TRAF protein (Rothe et al., 1995a; Cao et al.,
1996; Takeuchi et al., 1996). Deletion of the RING finger
domain of TRAF2results ina truncated, dominant-nega-
tive mutant protein that interferes with the downstream
signaling induced by stimulation of the TNF receptors
and CD40 (Rothe et al., 1995a).
It has been shown that TRAF2 transduces signals
required for TNF-mediated activation of the transcrip-
tion factor NF-?B (Rothe et al., 1995a) and the stress-
activated protein kinase (SAPK or c-J un N-terminal ki-
nase [J NK]; Liu et al., 1996; Natoli et al., 1997; Reinhard
et al., 1997). The precise role of J NK/SAPK in pro-
grammed cell death is still controversial. Studies using
cells withatargeted disruptionofSEK1,a kinasedirectly
upstream of J NK/SAPK, have demonstrated that this
kinase pathway plays a protective role against various
cytotoxic stimuli (Nishina et al., 1997). It has also been
recently established that TNF-mediated NF-?B activa-
tion is important for establishing cellular resistance to
TNF-induced apoptosis (Beg and Baltimore, 1996; Liu
et al., 1996; VanAntwerp et al., 1996; Wang et al., 1996).
Blockade of NF-?B activation either by knockout of the
p65gene orby overexpressionof thesuperinhibitorI?B?
renders cells sensitive toTNF-induced celldeath.TRAF2
is also required forthe recruitment of c-IAP1 and c-IAP2
to the TNFR complexes (Rotheet al.,1995b). Sincethese
TRAF2 activities are anti-apoptotic in nature, TRAF2
may play a key role in regulating the well-known pro-
apoptotic activity of TNF.
In this study we used gene targeting to generate a
mutantmouse strainspecifically deficientinTRAF2.Sur-
prisingly, alack ofTRAF2 leads toembryonic and neona-
tal lethality. In addition, we demonstrate that while
TRAF2 is required for the activation of the J NK/SAPK
pathway, signaling via NF-?B can occur in the absence
of TRAF2. TNF signaling in traf2?/?cells was found to
Wen-Chen Yeh,*²Arda Shahinian,*²Daniel Speiser,²
J anine Kraunus,*Filio Billia,²Andrew Wakeham,*²
J ose ÂLuis de la Pompa,*²David Ferrick,*?Betty Hum,*
Norman Iscove,²Pamela Ohashi,²Mike Rothe,³
David V. Goeddel,³and Tak Wah Mak*² §
²Department of Medical Biophysics
Department of Immunology
University of Toronto
Ontario Cancer Institute
Canada M5G 2C1
2 Corporate Drive
South San Francisco, California 94080
TRAF2 is an intracellular signal-transducing protein
recruited to the TNFR1 and TNFR2 receptors following
TNF stimulation. To investigate the physiological role
mice appeared normal at birth but became progres-
sively runted anddied prematurely. Atrophy of the thy-
mus andspleenand depletionofB cellprecursors also
were observed. Thymocytes and other hematopoietic
progenitors were highly sensitive to TNF-induced cell
death and serum TNF levels were elevated in these
TRAF2-deficient animals. Examination of traf2?/?cells
revealed a severe reduction in TNF-mediated J NK/
SAPK activation but a mild effect on NF-?B activation.
These results suggest that TRAF2-independent path-
ways of NF-?B activation exist and that TRAF2 is re-
quired for an NF-?B±independent signal that protects
against TNF-induced apoptosis.
Tumornecrosis factor(TNF)is a multifunctionalcytokine
that plays an important role in the regulation of the im-
mune and inflammatory responses (Fiers, 1991). Exces-
sive TNF also contributes to pathological conditions
such as septic shock (Beutler and Cerami, 1988). The
many actions of TNF are mediated by two cell surface
receptors, TNF receptor-1 and TNF receptor-2 (TNFR1
and TNFR2), which are present on most cell types (Tar-
taglia and Goeddel, 1992; Vandenabeele et al., 1995).
TNF binding induces receptor aggregation, resulting in
the recruitment of a number of cytoplasmic signaling
proteins to the two distinct TNFR complexes (Rothe et
al., 1994; Hsu et al., 1995, 1996a, 1996b; Rothe et al.,
1995b; Shuet al., 1996).One of these signaling proteins,
TNFR-associated factor-2 (TRAF2), is a component of
§To whom correspondence should be addressed (e-mail: tmak@
?Presentaddress: Rigel, Inc., 772 Lucerne Drive, Sunnyvale, Califor-
Figure 1. Targeted Disruption of the Murine traf2 Gene
(A) A portion of the traf2 endogenous locus containing 6 exons (boxes). (Top) TRAF2 cDNA. The domains encoded by the represented exons
are indicated. The targeting construct (T-TRAF2) was designed to replace part of the two exons that encode the ring fingerdomain of TRAF2
with a PGK-neogene cassette. (Bottom)The mutantlocus resulting fromhomologous recombination.An extra BamHIrestrictionsite introduced
by the PGK-neo cassette was used for the diagnosis of homologous recombination.
(B) Southern blot analysis of tail DNA harvested from wild-type, traf2?/?, and traf2?/?animals. Tail DNA was digested with BamHI and probed
with the radiolabeled flanking probe (Figure 1A).
(C and D)Western blotanalysis of TRAF2protein expression. Total liverlysate, 800 ?g, from wild-type, traf2?/?, and traf2?/?mice was subjected
to partial purification using glutathione beads saturated with a GST-TNFR2 fusion protein (Rothe et al., 1994). A sample of beads with GST
protein alone was used as a control. The partially purified lysates were fractionated on a 10% denatured polyacrylamide gel, and the Western
blot was probed with polyclonal antibodies raised against a peptide derived from the TRAF-N domain (C) or a peptide from the TRAF-C
domain of TRAF2 (D).
be defective and could be linked to at least some of the
observed mutant phenotypes. The results presented in
this paper clarify the proposed roles of TRAF2 in TNF
signaling and provide important insights into the under-
lying molecular mechanisms of TNF-mediated cellular
life and death.
Southern blot analysis (Figure 1B). TRAF2 protein ex-
pression in tissues was determined by affinity purifica-
tionand subsequent Western blot analysis, which dem-
onstrated the absence of intact or truncated TRAF2 in
traf2?/?animals (Figures 1C and 1D). Initial examination
of the progeny from traf2?/?parents in the 129 ? B6
background revealed a reduced frequency of traf2?/?
births. Of the first 310 live pups examined, only 6.5%
were traf2?/?. As shown in Table 1, examination of the
embryos from heterozygous traf2 matings revealed a
normal Mendelian ratio of TRAF2 mutants until embry-
onic day 14.5 (E14.5). No developmental ormorphologi-
cal abnormalities were observed in mutant embryos at
this stage. The frequency of viable traf2?/?offspring in-
creased when the genetic background was mixed (B6?
BALB/c, 10.0%), and decreased when the background
was closer to the inbred B6 strain (3.4%). This observa-
tion suggests that the B6 inbred strain may have a ge-
netic defect that interacts with the null mutation of
TRAF2 to produce increased lethality.
Regardless of strain background, viable traf2?/?mice
Disruption of TRAF2 in Mice Results in Runtism,
Wasting, and Premature Lethality
To elucidate the functionof TRAF2 in vivo, we disrupted
the murine traf2 gene in embryonic stem (ES) cells with
a targeting vector T-TRAF2 (Figure 1A). Heterozygous
ES cell lines containing a mutated traf2 allele were in-
jected into C57BL/6 (B6)blastocysts and male chimeras
with germline transmission were used to generate
traf2?/?animals. Heterozygous animals were further
backcrossed to B6 mice and at the same time were
interbred to obtain homozygosity. Homologous recom-
bination and genotype screening were confirmed by
TNF-Induced Cell Death in TRAF2-Deficient Mice
Table 1. Genotypic Analyses of the Progeny from TRAF2 Heterozygous Intercrosses
Number of the Genotypes
Timed breedings were set up between traf2?/?males and females (129 ? B6 background). E14.5 was defined as 14.5 days after fertilization.
Tails of newborns were collected at the time of birth; tails of live-born pups were collected about 3 weeks after birth or at the time of death.
The genotypes were determined primarily by PCR analyses, as described in Experimental Procedures. Early genotyping was confirmed by
Southern blot analysis.
appeared normal at birth but became smaller than their
littermates after 2±3 days. traf2?/?mutants developed
normally in their four extremities, but their body trunks
were devoid offat deposits and reducedin musclemass
compared to those of their wild-type littermates. The
mutantanimals became more runted withtime, and pre-
mature mortality was observed, with the incidence of
death peaking at the age of 10±14 days. At the time of
death, the body weights of mutant animals ranged from
20% to 50% of normalcontrols. Only a smallpercentage
of traf2?/?mice survived longerthan3 weeks, and these
were typically eitherfrom a small litterorfromthe hybrid
background. Most organs of traf2?/?animals, while pro-
portionally smaller than those of control littermates, re-
vealed no gross developmental abnormalities. The ex-
ceptions were the thymus and spleen, which were
extremely atrophic. Microscopic examination of tissues
of traf2?/?animals revealed lymphoid depletion in the
thymus and spleen and lack of secondary germinal
centers inspleenand lymphnodes. Examinationofperi-
pheralblood cells frommutantanimals showed lympho-
penia of bothT and B lymphocytes and relative granulo-
cytosis, but normal erythrocyte counts.
TRAF2-Deficient Hematopoietic Progenitors
Are Sensitive to TNF-Induced Cell Death
To investigate the defects of hematopoietic lineages
in traf2?/?animals, cells of the bone marrow, thymus,
spleen, and lymph nodes were harvested and analyzed
by cell counts and flow cytometry. No obvious abnor-
malities were observed in the erythroid or myeloid lin-
eages. Thymocytes from traf2?/?animals were reduced
in totalcellnumberand were devoid ofCD4?CD8?(dou-
ble-positive [DP]) cells as the mutant mice approached
the terminal stage (data not shown). However, thymo-
cytes derivedfromyoungeranimals orfetalthymic organ
cultures showed a profile of surface markers similar
to the wild type (data not shown), indicating normal
thymocyte development in the absence of TRAF2. In
bone marrow, B cell precursors (B220?CD43?) from
traf2?/?animals were reduced compared to normal con-
trols (Figure 2), buta complete blockade of B cell devel-
opment was not observed.
The depletionof lymphocytes intraf2?/?animals could
be due to a pathological up-regulation of cytotoxic fac-
tors such as steroid hormones or to an intrinsic defect
of cell survival, or both. Because TRAF2 is involved in
Figure 2. B Cell Development in the Bone
Marrow of traf2?/?Survivors
Bone marrow cells were harvested from a
9-day-old traf2?/?mutant and its littermates
and were subjected to flow cytometric analy-
sis as described inExperimental Procedures.
The numbers shown in each quadrant repre-
sent the percentage of total viable cells.
TNF signaling, which, through TNFR1, includes both
pro-apoptotic and anti-apoptotic pathways, we set out
to determine the exactrole of TRAF2 in these pathways.
Thymocytes from traf2?/?and control animals were cul-
tured at 37?C and treated with the cytotoxic agents
mouse TNF?(mTNF?), anti-Fas antibody, ordexametha-
sone. After20hr, cells were harvested and triple-stained
with CD4, CD8, and 7-aminoactinomycin D (7-AAD). As
shown in Figures 3A and 3B, relative to untreated con-
trols, DP cells from traf2?/?mice were more sensitive
to TNF-induced cell death than were cells from either
traf2?/?or traf2?/?mice. In contrast, the viability of DP
cells of allgenotypes was reduced to a similardegree by
Fas signaling ordexamethasonetreatment.Comparison
of untreated traf2?/?and traf2?/?cells revealed in-
creased spontaneous apoptosis of traf2?/?DP thymo-
cytes over a 20 hr period of culture at 37?C (Figure 3B),
even though the starting populations of both genotypes
contained the same percentage of viable DP cells. Simi-
lar results were obtained when the rate of cell death of
fetal thymocytes harvested from thymic organ cultures
from E14.5 embryos was examined (data not shown).
OtherTRAF2-deficient hematopoietic precursors also
were tested for their sensitivity to TNF-induced death.
In vitro colony-forming assays were performed using
fetal liver cells from E14.5 embryos in the presence or
absence of 10 ng/ml mTNF?. When plated in the ab-
sence of TNF?, erythroid (CFU-E and CFU-E/MEG) and
myeloid (CFU-G, CFU-GM, and CFU-MAC) colonies
from traf2?/?fetal livers developed normally. In the
presence of TNF?, however, very few or no colonies
developed from traf2?/?fetal livers, while colonies from
traf2?/?or traf2?/?fetal livers were only mildly sup-
pressed (Table 2).
To assess further the viability of TRAF2-deficient he-
matopoietic cells in vivo, we performed adoptive trans-
fer experiments by injecting E14.5 embryonic liver cells
into lethally irradiated B6 mice. Wild-type and traf2?/?
fetal livercells were capable of reconstituting the hema-
topoietic system, and 100% survival rate (6 of 6 for ?/?
and 19 of 19 for ?/?) was observed in the recipients
when examined 3 months after transfer. In contrast, of
the irradiated B6 hosts receiving traf2?/?fetal liver cells,
only 45% (5 of 11) survived past the first 3 weeks after
transfer. The surviving traf2?/?fetal liverrecipients were
further examined 3 months after transfer, and poorly
reconstituted hematopoiesis including anemia and lym-
phopenia was observed (data not shown).
Figure 3. The Sensitivity of TRAF2-Deficient Thymocytes to Murine
(A)Survival of CD4?CD8?(DP) thymocytes aftervarious treatments.
Thymocytes from a relatively healthy 2-week-old traf2?/?mouse and
its littermates were harvested and cultured at 37?C for 20 hr in
the absence or presence of the indicated stimuli. After treatment,
thymocytes were triple-stained with CD4, CD8, and 7-AAD. Viable
DP cells (negative 7-AAD stain) from individual treatments were
plottedas percentagesof the totalviableDP cells from the untreated
controls ofthe respective genotypes. Eachtreatment group is com-
posed of three bars with standard errors, representing traf2?/?,
traf2?/?, and traf2?/?DP thymocytes.
(B) Flow cytometric analysis of viable DP thymocytes. traf2?/?or
traf2?/?thymocytes untreated or treated with 10 ng/ml mTNF were
triple-stained as above and gated on cells with negative 7-AAD
stain. The number at the top right of each panel is the percentage
of total cells that were viable DP cells.
Defective TNF Signaling in TRAF2-Deficient
Mice Is Mediated by TNFR1
To dissect the TNF signaling defects in traf2?/?thymo-
cytes, human TNF (which interacts with mouse TNFR1,
but not mouse TNFR2) and receptor-specific agonist
antibodies against TNFR1 or TNFR2 were examined in
the presenceofthe proteinsynthesis inhibitorcyclohexi-
mide. Thymocytes from traf2?/?mice were more sensi-
tive than traf2?/?or traf2?/?cells to human TNF± or
anti-TNFR1±induced cell death (Figure 4). Anti-TNFR2
treatment did not kill cells of any genotype. Similar re-
sults were obtained in experiments using fetal thymo-
cytes (data notshown).These observations indicatethat
TRAF2 plays animportant role in protecting thymocytes
from TNF-induced apoptosis and that in the absence of
TRAF2, cells become susceptible to TNF-induced cell
death mediated through TNFR1.
To determine whether TNF was involved in the re-
duced viability of traf2?/?mice, TNF levels in sera from
animals aged 2±3 weeks were examined. Whereas TNF
TNF-Induced Cell Death in TRAF2-Deficient Mice
Table 2. In Vitro Colony-Forming Assay of Fetal Liver Cells
GenotypeTreatmentE E/MEGG MAC GMMulti TOTAL
36.0 ? 1.2
34.5 ? 5.8
8.5 ? 2.3
15.3 ? 6.0
39.0 ? 6.3
12.8 ? 1.9
82.0 ? 8.7
79.8 ? 4.3
37.3 ? 4.0
29.3 ? 7.4
4.5 ? 1.6
4.5 ? 1.7
206.3 ? 11.1
176.0 ? 16.9
32.6 ? 4.1
31.8 ? 3.6
8.4 ? 1.6
9.3 ? 1.6
38.6 ? 4.4
11.1 ? 1.4
80.0 ? 5.9
60.2 ? 9.0
20.8 ? 2.4
13.6 ? 2.7
3.1 ? 0.5
1.7 ? 0.4
186.5 ? 14.9
125.6 ? 17.4
22.8 ? 5.0
0.0 ? 0.0
14.3 ? 4.6
0.25 ? 0.25
11.0 ? 3.2
1.25 ? 0.75
110.0 ? 5.8
5.5 ? 2.8
20.3 ? 5.4
0.5 ? 0.5
4.3 ? 1.0
0.0 ? 0.0
182.5 ? 22.6
7.5 ? 3.5
Fetal liver cells of E14.5 embryos were plated in media facilitating the growth and differentiation of hematopoietic colonies in the absence or
presence of 10 ng/ml recombinant mTNF?, as described in Experimental Procedures. Numbers in each column represent colonies of specific
or mixed lineages identified per 105fetal liver cells. Only colonies with more than 50 cells were counted.
E, erythroid colony; E/MEG, erythroid and megakaryocytic colony; G, granulocytic colony; MAC, macrophage colony; GM, granulocytic and
macrophage colony; multi, multilineage colony (containing more than two lineages).
could notbe detected insera from wild-typeand hetero-
zygous animals, TRAF2-deficient animals showed clearly
detectable levels of serumTNF (Figure 5). The elevation
of serum TNF could be either an indirect consequence
of illness or, alternatively, a direct effect of the TRAF2
deficiency that results in dysregulated TNF synthesis.
In any case, this result demonstrates in vivo the signifi-
cance of the supersensitivity of TRAF2-deficient cells
in the absence of TRAF2, several independent TRAF2-
deficient and control embryonic fibroblast (EF) lines
were derived. The EF cell lines were stimulated with
mTNF for various periods, and NF-?B activation was
determined by gel mobility shift assays. Surprisingly,
near-normal NF-?B activation was clearly demonstrable
in traf2?/?EF cells, albeit with delayed kinetics (Figure
6A). By 90 min after mTNF treatment, the proportion of
activated NF-?B in traf2?/?EF cells was equivalent to
the proportion present incontrol EF cells. Similardelays
were observed in other traf2?/?EF lines as well as in
position of activated NF-?B in traf2?/?cells was exam-
ined using specific antibodies to different NF-?B family
members, including p65 (RelA), p50, and c-Rel (Figure
6B). No obvious differences were observed in the su-
pershiftpatternbetweenwild-typeand mutantcells (Fig-
The activation of J NK/SAPK by mTNF was also exam-
ined in traf2?/?EF cells. The activation of J NK/SAPK in
wild-type cells peaked at about 10±15 min after mTNF
treatment (Figure 6C). However, cells lacking TRAF2
showed greatly reduced J NK/SAPK activity. This defect
in J NK/SAPK activation was specific to the TNF signal-
ing pathway, because other stress signals, such as an-
isomycin (Figure 6C), ultraviolet irradiation, and heat
TRAF2 Is Required for TNF-Mediated J NK/SAPK
Activation and the Early Phase
of the NF-?B Response
TRAF2 has beenimplicated in the TNF-mediated activa-
tion of the NF-?B and J NK/SAPK pathways. To test
whether the activation of these pathways is abrogated
Figure 4. TNFR1 versus TNFR2 Signaling in the TNF Hypersensitiv-
ity of TRAF2-Deficient Cells
Thymocytes from a 2-week-old traf2?/?mutant and its littermates
were cultured inthe presenceof cycloheximide(C) and eithermTNF,
human TNF (hTNF), murine TNFR1 agonist antibody (anti-TNFR1),
or TNFR2 agonist antibody (anti-TNFR2). After incubation for 20 hr,
cells were triple-stained with CD4, CD8, and 7-AAD. The results
are shown as viable DP cells remaining after various composite
treatments, expressed as a percentage of the viable DP cells surviv-
ing cycloheximide treatment alone for each genotype.
Figure 5. Serum TNF Levels in Surviving TRAF2-Deficient Mice
Serum TNF concentrations were determined in three pairs of traf2?/?
mice and their littermate controls (?/?, ?/?; CC9, CC11; and WL,
WS), as describedin ExperimentalProcedures. Micewere sacrificed
at the age of 2±3 weeks. nd, not detectable.
Figure 7. TNF Cytotoxicity of traf2-/-and traf2?/?EF Cells
TRAF2-deficient and wild-type EFs were treated with mTNF? (10
ng/ml)andcycloheximide (0.25?g/ml)forthe indicatedtimeperiods.
Cells viability was determined by trypan blue exclusion. Two inde-
pendent cell lines of each genotype are shown.
dramatically reduced (20%±40% viable) inthe presence
of mTNF?for8 hr(Beg and Baltimore, 1996). Incompari-
son, we examined the sensitivity of traf2?/?EFs to
mTNF? (10 or 100 ng/ml), and after 24 hr of treatment
cell viability remained approximately 90%. This indi-
cated that there are differential sensitivities to TNF
among different TRAF2-deficient cell types. We then
treated TRAF2-deficient and control EFs with mTNF?
(10 ng/ml)and cycloheximide (0.25 ?g/ml)and analyzed
cell viability after 6 or 24 hr. While wild-type EFs were
not affected under this condition, TRAF2-deficient EFs
started to die as early as 6 hr after treatment, and only
15%±40% of the mutant cells survived by 24 hr after
treatment (Figure 7). Taken together, these results sug-
gest that TRAF2 is important in mediating a protective
signal parallel but complementary to NF-?B-mediated
Figure 6. TNF-MediatedNF-?B andJ NK/SAPK ActivationinTRAF2-
(A) TRAF2-deficient EF cells and wild-type controls were incubated
with mTNF (10ng/ml)forthe indicatedtimeperiods.Nuclearextracts
(10 ?g) harvested at each time point were incubated with a radiola-
beled probe containing NF-?B binding sites, and NF-?B activation
was determined using a gel mobility shift assay as described in
(B) TRAF2-deficientfibroblasts and wild-typecells were treatedwith
mTNF for 30 min. Nuclear extracts (10 ?g) were incubated with
specific antibodies against different members of the NF-?B family
for 30 min before the addition of the radiolabeled probe and gel
mobility shift assay.
(C) TRAF2-deficient fibroblasts and wild-type cells were incubated
with TNF for various times as indicated, or with anisomycin (aniso),
cycloheximide (cyclo),orcycloheximide plus TNF for 30min. Protein
lysates from individual samples were immunoprecipitated with a
J NK/SAPK-specific antibody, and kinaseactivitywas assayed using
a GST-c-J un fusionprotein as the substrate, as described in Experi-
Ex vivo experiments have demonstrated that TRAF2
plays an important role in transducing signals elicited
by multiplemembers of the TNFR superfamily,including
TNFR1 (Hsu et al., 1996a; Liu et al., 1996; Shu et al.,
1996), TNFR2 and CD40 (Rothe et al., 1995a), CD30
(Ansieau et al., 1996; Gedrich et al., 1996; Lee et al.,
1996; Aizawa et al., 1997; Boucheret al., 1997; Duckett
et al., 1997; Tsitsikov et al., 1997), and herpesvirus entry
mediator (Montgomery et al., 1996; Hsu et al., 1997;
Marsters et al., 1997). In the present investigation, we
used gene targeting to show in vivo that TRAF2 defi-
ciency produces mice thatfailto thrive,are runted, show
lymphopenia, and die prematurely. This severe pheno-
type contrasts sharply with the milder phenotypes of
TNFR1 and TNFR2 knockout mice (Pfeffer et al., 1993;
Ericksonet al., 1994), which generally are ingood health
in a pathogen-free environment. Mice withtargeted dis-
ruptions of CD40 and CD30 have also shown limited
phenotypes withdefects specific to the immune system
(Kawabe et al., 1994; Amakawa et al., 1996). However,
mice deficient in TRAF3, a protein homologous to
TRAF2, are also runted, exhibit lymphopenia, and die at
shock (data not shown), were still able to activate J NK/
SAPK in traf2?/?cells. Finally, we examined c-IAP1 and
c-IAP2 andfoundthat theywere expressed atequivalent
levels in TRAF2-deficient and wild-type cells (data not
TRAF2-Deficient EF Cells Are More Sensitive
to TNF-Induced Cell Death in the
Presence of Cycloheximide
Wild-type EF cells are resistant to TNF treatment, while
the viability of NF-?B p65?/?(RelA-deficient) EFs are
TNF-Induced Cell Death in TRAF2-Deficient Mice
an early postnatal stage (Xu et al., 1996). Such a severe
phenotype cannot be explained by the findings to date
that TRAF3 specifically associates with CD40 and the
Epstein-Barr virus latent membrane protein 1 (Hu et al.,
1994; Cheng et al., 1995; Mosialos et al., 1995; Sato et
al., 1995; Devergne et al., 1996; Izumi et al., 1997). We
conclude that TRAF2 and TRAF3 are probably involved
inimportant transductionpathways, otherthan TNF sig-
naling, that remain to be identified.
The lymphoid depletion common to TRAF2- and
TRAF3-deficient mice could be either the result of an
intrinsic defect in cell survival or simply a consequence
of the up-regulation of cytotoxic factors such as steroid
hormones. Since TRAF3-deficient fetal liver cells were
able to reconstitute allhematopoietic lineages in lethally
irradiated recipients, it was concluded that TRAF3 is not
essential forhematopoietic cell survival(Xuet al., 1996).
However, our examinations of TRAF2-deficient cells re-
vealed the existence of intrinsic defects inthe response
of these cells to TNF.Thymocytes and otherhematopoi-
etic progenitors from traf2?/?animals and embryos were
highly sensitive to TNF-induced cell death. Adoptive
transferexperiments showed that traf2?/?fetallivercells
were unable to reconstitute the hematopoietic compart-
ment of lethally irradiated mice. In addition, serum TNF
was readily detectable in the TRAF2 mutant animals.
These results suggest that both intrinsic cell defects
and the elevation of a cytotoxic factor contribute to
the phenotype of the TRAF2 knockout mouse, and that
TRAF2 and TRAF3 play distinctly different roles in intra-
Some aspects of the TRAF2 mutant phenotype are
consistent with the toxicity induced by excessive levels
of TNF (Beutler and Cerami, 1988; Probert et al., 1993).
The TRAF2-deficient mice are devoid of fat deposits
and muscle mass, and wasting starts shortly after birth
and progresses until the time of death. In addition, thy-
mocytes and B cell precursors are progressively de-
pleted. However, we did not observe in TRAF2-deficient
mice the histopathological changes, including vascular
thrombosis and extensive tissue necrosis, reported in
transgenic mice that express high levels (100±1300
pg/ml in serum) of human TNF (Probert et al., 1993).
Indeed, comparatively moderate serum TNF levels (20±
50 pg/ml) were detected in TRAF2-deficient animals. It
is possible that the persistently detectable serum TNF
is derived from dysregulation of TNF synthesis and se-
cretion in animals lacking TRAF2. Further investigation
into the molecular mechanisms regulating TNF gene
expression and processing will help address this issue.
Could the complete phenotypes of TRAF2-deficient
mutants be explained by TNF toxicity? Becausethe em-
bryonic lethality is the prominent feature of traf2?/?mu-
tants in the genetic background close to the inbred B6
strain, we attempted to rescue the mutant embryos by
TNF inhibition using a specific anti-TNF antibody. We
injected anti-TNF monoclonal antibody (8 ?g/g body
ster immunoglobulins into the peritoneum of pregnant
females from heterozygous intercrosses on day 11.5 of
gestation. These heterozygous animals were generated
by repeated backcrossing to B6 at least three times. Of
eightlitters fromthe controlinjection, 56pups were born
and only 1 (1.8%) was traf2?/?. From 15 females treated
with anti-TNF antibody injection, 105 pups were born
and 10 (9.5%) were identified as traf2?/?. Thus, there
was a clearbut incomplete rescue of the TRAF2 mutant
embryos by TNF inhibition. This rescue effect was not
enhanced by increasing the frequency of the injections
(data not shown). Mild improvementbutincomplete res-
cue was also observed in the live-born TRAF2 pups
regularly injected with the anti-TNF antibody (data not
shown). However, the small number of overall traf2?/?
live-born survivors is a limitation; in future studies, the
role of TNF toxicity in traf2?/?mutants could be ad-
dressed more completely by generation of TRAF2 and
TNF double-knockout animals.
This studyalso shows thatthe defective TNF signaling
in TRAF2-deficient cells that leads to increased cell
death is mediated by TNFR1. Previous studies have
shown that TRADD, an adaptor protein that binds di-
rectly to the death domain of TNFR1, can transduce
signals bothforapoptosis and forNF-?B activation (Hsu
et al., 1995). TRADD achieves these responses by re-
cruiting eitherFADD (Fas-associated deathdomainpro-
tein) orTRAF2 to the TNFR1 complex, thereby signaling
apoptosis or NF-?B activation, respectively (Hsu et al.,
1996a). Activated NF-?B in turn generates a protective
signal against TNF-induced apoptosis (Beg and Balti-
more, 1996; Liu et al., 1996; Van Antwerp et al., 1996;
Wang et al., 1996). Thus, in the absence of TRAF2, TNF
signaling through TNFR1 is mediated primarily by a
TRADD±FADD interaction, which leads to apoptosis.
This signal shunting suggests that in normal cells, there
exists animportant balance of signaling initiated by TNF
and that TRAF2 is essential for the protective signaling
Examinationof TNF-inducedNF-?B activationinTRAF2-
deficient cells revealed unexpectedly normal levels of
this transcription factor complex, although its kinetics
of activation were delayed. Thus, we have shown that
NF-?B can be activated by TNF even in the absence of
TRAF2. Three of six TRAF proteins characterizedto date
(TRAF2, TRAF5, and TRAF6) are able to mediate the
activation of NF-?B (Rothe etal., 1995a; Cao et al., 1996;
Ishida et al., 1996a, 1996b; Nakano et al., 1996; Aizawa
et al., 1997). TRAF5, a protein similarto TRAF2, is widely
expressed invarious tissues and is alsoused by multiple
members of the TNF receptor superfamily, including
CD40, CD30, and the lymphotoxin-? receptor (Nakano
et al., 1996). It is possible that TRAF5 can substitute for
TRAF2 in NF-?B activation induced by TNF or other
TNF-related factor signals, consistent with the ability of
TRAF5 to interact with TRADD (D. V. G., unpublished
data). Alternatively, RIP (receptor-interacting protein),
another signaling protein that interacts with TRADD,
TNFR1, and TRAF2, may be moreimportant thanTRAF2
in activating NF-?B in vivo (Stanger et al., 1995; Hsu
et al., 1996b). Consistent with this hypothesis, a RIP-
deficient mutant J urkat cell line is defective in TNF-
induced NF-?B activation but fully susceptible to Fas-
induced apoptosis (Ting et al., 1996).
An obvious question remaining to be addressed is the
molecular mechanism(s) by which TRAF2 transmits its
protective signal. One possibility is thatthe transmission
occurs through preformed complexes between TRAF2
and c-IAP1 and c-IAP2 (Rothe et al. 1995b; Shu et al.,
1996). The c-IAPs have been proposed to signal an anti-
apoptotic function, but the precise mechanism remains
Western Blot Analysis
Total liverlysates from traf2?/?, traf2?/-, and traf2-/-littermates were
homogenized in lysis buffer containing 50 mM Tris-Cl (pH 7.4), 100
mM NaCl, 1% Nonidet P-40 (NP-40), 2 mM EDTA, 10% glycerol,
and the protease inhibitors aprotinin (1 ?g/ml), leupeptin (1 ?g/ml),
and phenylmethylsulfonyl fluoride (PMSF, 10 ?M/ml). Next, 800 ?g
lysate from each genotype was partially affinity purified for TRAF2
using glutathione beads saturated with a fusion protein of glutathi-
one S-transferase (GST)±TNFR2 (Rothe et al., 1994). After five
washes in lysis buffer containing 250 mM NaCl, the protein samples
were fractionated by 10% SDS±polyacrylamide gel electrophoresis
and blotted onto a nitrocellulose membrane. The Western blot was
probed with a polyclonal antibody against a specific polypeptide
derivedfrom the TRAF2±TRAF-Ndomainorwith anantibody against
a polypeptide from the TRAF2±TRAF-C domain. The blot was then
developed using an enhanced chemiluminescence system ac-
cording to the manufacturer's instructions (Amersham).
unknown. TRAF5, despite its ability to activate NF-?B,
cannot interactwithorrecruit cIAPs to eitherTNF recep-
tor(D.V.G., unpublisheddata).This leaves opena possi-
ble specific role for TRAF2 in the recruitment of c-IAPs
in TNF-mediated cell protection. Alternatively, TRAF2-
mediated protection may occur through the TNF-induced
J NK/SAPK pathway, whichis greatlyreduced inTRAF2-
deficient cells. Recentstudies have dissociatedthe acti-
vation of J NK/SAPK by TNF from the induction of
apoptosis (Liuet al., 1996; Natoliet al., 1997;Reinhard et
al., 1997), although the phenotype of the SEK1 knockout
mouse suggests that this pathway may be involved in
the regulation of apoptosis (Nishina et al. 1997).
One intriguing hypothesis possibly accounting for a
protective effect of TRAF2 suggests that ªsurvival ef-
fectorº genes that have yetto be identified may be tran-
scriptionally activated by NF-?B. In this scenario,
TRAF2-deficient cells undergo TNF-induced apoptosis
because of delayed activation of NF-?B. It is possible
that the early phase of NF-?B activation is crucial to
induce the expressionof the survivaleffectors properly.
Alternatively, perhaps TRAF2 induces an NF-?B±inde-
Such a signal might then collaborate with NF-?B at the
transcriptional level or interact directly with the survival
effectors. Further identification of these downstream
cell survival transducers will provide an important blue-
print of the molecular mechanism that antagonizes the
cell death machinery.
Flow Cytometric Analysis of TRAF2-Deficient
Single-cell suspensions were prepared from lymphoid organs, in-
cluding thymus and bone marrow, as previously described (Ama-
kawaetal.,1996). Forantibodydetectionineachcytometric analysis
5 ? 105cells were used The antibodies used in this study, including
phycoerythrin(PE)±conjugatedanti-CD8? and anti-B220antibodies,
and fluorescein isothiocyanate (FITC)±conjugated anti-CD4, anti-
CD43, and anti-IgM antibodies were obtained from Pharmingen.
Afterstaining, fluorescent signals onthe cellsurfaces were analyzed
with a FACScan flow cytometer (Becton Dickinson) and CellQuest
Thymocyte Death Assay
Freshly isolated thymocytes were cultured in RPMI medium con-
taining 5% fetal bovine serum and were plated at 5 ? 105cells/ml
in eachwell of a 24-well dish. Various cytotoxic agents were added,
including dexamethasone 1 ?M, mTNF 10 ng/ml, anti-Fas antibody
(J o-2 clone, a gift from S. Nagata) 1 ?g/ml, human TNF 10 ng/ml,
cycloheximide 50 ?g/ml, cycloheximide plus mTNF, cycloheximide
plus human TNF, cycloheximide plus anti-TNFR1 agonist antibody
(1:300 dilution), and cycloheximide plus anti-TNFR2 agonist anti-
body (1:300 dilution). After 20 hr of incubation at 37?C in 5% CO2,
the cells were harvested, stained with PE-CD8?, FITC-CD4, and
7-AAD (used at 1 ?g/ml, Sigma), and analyzed with a FACSscan
cytometerand CellQuest software as previously described (Nishina
et al., 1997). 7-AAD is a vital dye which enters dead cells after loss
ofmembrane integrity. It allows quantitation of dead and viable cells
by the FL-3 channel on a FACScan cytometer and can be used in
combination with FITC-conjugated (channel FL-1) and PE-conju-
gated (channel FL-2) antibodies against specific cell surface mark-
ers (Schmid et al., 1994). Viable double-positive (CD4?CD8?) cells
were displayedona dotplot gatedonthymocytes negatively stained
by 7-AAD. Individual treatments were measured in triplicate, and at
least three independent experiments were performed. Induction of
thymocyteapoptosis was confirmed byDNA laddering as previously
described (Amakawa et al., 1996). Parallel tryptan-bluevital staining
was also performed.
Generation of traf2?/?Mice by Gene Targeting
A genomic DNA clone fortraf2 was isolated byscreening a129/Sv/J
mouse genomic DNA library using a probe derived from the 5? end
of the mouse traf2 cDNA. The cDNA region encoding the TRAF2
ring finger domain was mapped to two exons separated by a 1.7
kb intron (Figure 1A). The targeting construct T-TRAF2 was made
by replacing the entire ring fingercoding region and the intervening
intron with the PGK-neo gene cassette in the reverse orientation to
the endogenoustraf2 gene. Linearized
transfected into ES cells (E14 clone, derived from 129/Ola mouse
embryos) by electroporation. Neomycin-resistant ES clones were
selected by 300 ?g/ml G418 (Sigma)in Dulbecco's modified Eagle's
medium supplemented with 15% fetal bovine serum and leukemia
inhibitory factor.Homologous recombinants were identifiedbypoly-
merase chain reaction (PCR) using a traf2-specific primer down-
stream of the targeting construct (5?-ATA CAC TTG CAC AGA CAT
ACA TGC AAG CAA-3?) and a primer in the PGK-neo gene cassette
(5?-AAG CGC ATG CTC CAG ACT GCC TTG GGA A-3?). Genomic
DNA was isolated from PCR-positiveclones and analyzed by South-
ern blot hybridization using the flanking probe (Figure 1A) and a
probe specific forthe neomycin-resistance gene.The flanking probe
was a radiolabeled 162 bp StuI±PstI fragment downstream of the
targeting construct. ES clones confirmed for homologous recombi-
nation and a single PGK-neo insertion were injected into C57BL/6J
blastocysts, which were then transferred into CD1 pseudopregnant
foster mothers. The resulting male chimeras were backcrossed with
C56BL/6J females, and germline transmission in F1 traf2?/?mice
was verified by Southern blot analysis. Heterozygous mice were
interbred to obtain traf2?/?mice. Genotyping of the F2 mice was
performed by PCR on tail genomic DNA and verified by Southern
blot analysis. The PCR primers for wild-type traf2 alleles were
5?-GGC TGA GCA GGC AGT GCT CAG AGA TTC-3? and 5?- CCT
CAT TCC GTT ACC AGT GTT ACA GA-3?. The primer pairs for the
mutantalleles were 5?-AAG CGC ATG CTC CAG ACT GCC TTG GGA
A-3? and 5?-CCT CAT TCC GTT ACC AGT GTT ACA GA-3?.
In Vitro Hematopoietic Colony-Forming Assay
The in vitro colony-forming assay was performed as previously de-
scribed (Karasuyama and Melchers, 1988). In brief, freshly isolated
fetal liver cells from E14.5 embryos were plated at a density of 105
cells/ml in a 3 cmdish. Cells were cultured in the Iscove's modified
Dulbecco's medium containing 1.2% methylcellulose, 4% fetal bo-
vine serum, 0.5% serum fraction V, 100 ?g/ml transferrin, and 10
?g/ml insulin. The culture medium was supplemented with the fol-
lowing cytokines: IL-1? (1 ng/ml, a gift from R. Shaw and J . J .
Mermod, Glaxo Research Institute, Geneva); IL-3 (15 U/ml, condi-
tioned medium of myeloma cells expressing murine IL-3, Karasu-
yama and Melchers, 1988); IL-11 (30 ng/ml, recombinant human IL-
11, Genetics Institute, Boston,MA);c-Kit ligand(3% v/v, conditioned
medium of Chinese hamster ovary cells expressing murine c-Kit
ligand, a gift from D. Donaldson, Genetics Institute, Boston); and
erythropoietin (1 U/ml). Plates treated with TNF received 10 ng/ml
TNF-Induced Cell Death in TRAF2-Deficient Mice
recombinant mTNF? (Genzyme). Erythroid colonies were counted
after 7 days, and granulocytic, macrophage, and megakaryocytic
colonies were counted after 12 days.
EF Death Assay
EF cells were growninDulbecco's modified Eagle's mediumsupple-
mented with 5% fetal bovine serum. Cells were plated at 2 ? 105
cells/ well of a six-well dish 2 days before the treatment. Recombi-
nant mTNF? (Genzyme) was used at 10 ng/ml, and cycloheximide
was used at250 ng/ml. The cells were incubated with medium only,
mTNF? alone, or mTNF? plus cycloheximide for 0, 6, or 24 hr. The
cells were then trypsinized and viability was determined by trypan
Measurement of Serum TNF
Whole blood was collected from traf2?/?mutants and control lit-
termates and centrifuged in a Microtainer serum separator (Becton
Dickinson), and the sera were stored at ?70?C. Measurement of
TNF was performed in duplicate on serum samples diluted 1:2 or
1:4 using an mTNF? enzyme-linked immunosorbent assay kit (Gen-
zyme) according to the manufacturer's instructions.
We thank Shigekazu Nagata for anti-Fas antibody (J o-2); J ames
Woodgett, J osef Penninger, and Hiroshi Nishina for reagents and
instructive discussions; J essie Xiong, Michelle Ng, and Michael Be-
zuly for excellent technical support; Bruce Patterson for pathology
consultation; and Petra Arck, Dave Clark, Louis-Martin Boucher,
Denise Bouchard, and members of T. W. M.'s laboratory for helpful
discussions and technical suggestions. We also thank Mary Saun-
ders for scientific editing and Irene Ng for administrative support.
This work was supported inpart by agrant fromthe NationalCancer
Institute of Canada. W.-C.Y. is a Medical Research Council of Can-
ada postdoctoral fellow.
Fetal Liver Translant
Eight-week-old C57BL/6-Ly 5.2 and C57BL/6/J mice were pur-
chased from Charles River Laboratory and J ackson Laboratory, re-
spectively. Recepients mice were preirradiated once by ?-ray for
1000 rad. About 3 million freshly isolated fetal livercells of all three
genotypes from E14.5 were injected intravenously into the recipi-
ents. After transfer, mice were treated with Clavulin-125F (1.25
mg/ml, SmithKline Beecham) for 2 weeks and monitored daily for
Gel Mobility Shift Assay
Nuclear extracts were harvested according to protocols previously
described (Pfefferet al.,1993). Inbrief, 2? 106cells, eitheruntreated
or stimulated with TNF for various times, were washed twice with
phosphate-buffered saline and resuspended in 400 ?l of buffer A
(10 mM HEPES [pH 7.8], 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA,
1 mM dithiothreitol, and 1 mM PMSF). After incubation on ice for 5
min, NP-40 was added to a final concentrationof 0.6%. Nucleiwere
pelleted and the cytoplasmic proteins were carefully removed. The
nuclei were then resuspended in buffer C (20 mM HEPES [pH 7.9],
0.4 M NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM dithiothreitol, and 1
mMPMSF).Aftervortexing andstirring for30 minat4?C, the samples
were centrifuged and the nuclear proteins in the supernatant were
transferredto a freshvial.Protein concentrations ofnuclearextracts
were determined by the Bio-Rad protein assay using bovine serum
albuminas the standard. Nuclear extract (10?g) was incubated with
an end-labeled, double-stranded, NF-?B-specific oligonucleotide
probe containing two tandemly positioned NF-?B-binding sites
(5?-ATC AGG GAC TTT CCG CTG GGG ACT TTC CG-3? and 5?-CGG
AAA GTC CCC AGC GGA AAG TCC CTG AT-3?). The reaction was
performed in a total of 20 ?l of binding buffer (5 mM HEPES [pH
7.8], 50 mM KCl, 0.5 mM dithiothreitol, 1 ?g poly [dI-dC], and 10%
glycerol) for 20 min at room temperature. For the supershift assay,
the nuclear extract was incubated with specific antibodies against
p65, c-Rel (Santa Cruz), or p50 (a gift from Dr. Z. Cao) for 30 min
beforeaddition ofthe labeled probe.Afterincubation, samples were
fractionated ona 5% polyacrylamide geland visualized by autoradi-
Received J uly 24, 1997; revised September 19, 1997.
Aizawa, S., Nakano, H.,Ishida, T., Horie, R., Nagai, M., Ito,K., Yagita,
H., Okumura,K., Inoue, J ., and Watanabe, T. (1997). Tumor necrosis
factor receptor-associated factor (TRAF) 5 and TRAF2 are involved
in CD30-mediated NFkappaB activation. J . Biol. Chem. 272, 2042±
Amakawa, R., Hakem, A., Kundig, T.M., Matsuyama, T., Simard,
J .J .L., Timms, E., Wakeham, A., Mittruecker, H.W., Griesser, H.,
Takimoto, H., et al. (1996). Impaired negative selection of T cells in
Hodgkin's disease antigen CD30-deficient mice. Cell 84, 551±562.
Ansieau, S., Scheffrahn,I., Mosialos,G., Brand,H., Duyster,J ., Kaye,
K., Harada, J ., Dougall, B., Hubinger, G., Kieff, E., etal. (1996). Tumor
necrosis factor receptor-associated factor (TRAF)-1, TRAF-2, and
TRAF-3 interact in vivo with the CD30 cytoplasmic domain; TRAF-2
mediates CD30-induced nuclear factor kappa B activation. Proc.
Natl. Acad. Sci. USA 93, 14053±14058.
Beg, A.A., and Baltimore, D. (1996). An essential role for NF-?B in
preventing TNF-?-induced cell death. Science 274, 782±784.
Beutler, B., and Cerami, A. (1988). Tumor necrosis, cachexia, shock
and inflammation: a common mediator. Annu. Rev. Biochem. 57,
Boucher, L.M., Marengere, L.E., Lu, Y., Thukral, S., and Mak, T.W.
(1997). Binding sites of cytoplasmic effectors TRAF1, 2, and 3 on
CD30 andothermembers ofthe TNF receptorsuperfamily.Biochem.
Biophys. Res. Commun. 233, 592±600.
Cao, Z., Xiong, J ., Takeuchi, M., Kurama, T., and Goeddel, D.V.
(1996). TRAF6 is a signal transducer for interleukin-1. Nature 383,
Cheng, G., Cleary, A.M., Ye, Z.S., Hong, D.I., Lederman, S., and
Baltimore, D. (1995). Involvement of CRAF1, a relative of TRAF, in
CD40 signaling. Science 267, 1494±1498.
Devergne, O., Hatzivassiliou, E., Izumi, K.M., Kaye, K.M., Kleijnen,
M.F., Kieff, E., and Mosialos, G. (1996). Association of TRAF1,
TRAF2, and TRAF3 with an Epstein-Barrvirus LMP1 domain impor-
tant for B-lymphocyte transformation: role in NF-?B activation. Mol.
Cell. Biol. 16, 7098±7108.
Duckett, C.S., Gedrich, R.W., Gilfillan, M.C., and Thompson, C.B.
(1997). Induction of nuclear factor kappaB by the CD30 receptor is
mediated by TRAF1 and TRAF2. Mol. Cell. Biol. 17, 1535±1542.
Erickson, S.L., de Sauvage, F.J ., Kikly, K., Carver-Moore, K., Pitts-
Meek, S., Gillett, N., Sheehan, K.C.F., Schreiber,R.D., Goeddel, D.V.,
and Moore, M.W. (1994). Decreased sensitivity to tumour-necrosis
factor but normal T-cell development in TNF receptor-2-deficient
mice. Nature 372, 560±563.
Fibroblasts (5 ? 105) were either untreated or treated with mTNF?
(100 ng/ml, various times),anisomycin(10?g/ml, 30min), cyclohexi-
mide (50 ?g/ml, 30 min), or cycloheximide plus mTNF? (30 min).
Cells were then lysed in an ice-cold lysis buffer (10 mM NaCl, 20
mM PIPES (Sigma) [pH 7.0], 0.5% NP-40, 5 mM EDTA, 0.05%
?-mercaptoethanol, 100 ?M Na3VO4, 50 mM NaF, 20 ?g/ml leupep-
tin, and 1 mM benzamidine). Cleared lysates were adjusted to equal
proteinconcentrations as determined by the Bio-Rad protein assay.
J NK/SAPK were immunoprecipitated using rabbit polyclonal anti-
J NK/SAPK IgG reactive against all J NK/SAPK isoforms (a gift from
Dr. J . Woodgett; Kyriakis et al., 1994), and harvested on protein
A-Sepharose beads. The immune complex was washed three times
with a phosphate buffercontaining 150 mM NaCl, 16 mM Na2HPO4,
4 mM NaH2PO4, 0.1% Triton X-100, 100 ?M Na3VO4, 50 mM NaF, 20
?g/ml leupeptin, and 1 mM benzamidine. The kinase assay was
performed in 20 ?l of kinase buffer (10 mM MgCl2, 50 mM Tris-HCl
[pH 7.5], 1 mM EGTA) in the presence of 1.2 ?Ci [32P]?-ATP and 5
?g GST-c-J un fusion protein (Kyriakis et al., 1994) for 30 min at
30?C. The samples were separated on a 12% SDS-polyacrylamide
gel and visualized by autoradiography.
Fiers, W. (1991). Tumor necrosis factor: characterization at the mo-
lecular, cellular and in vivo level. FEBS Lett. 285, 199±212.
Gedrich, R.W., Gilfillan, M.C., Duckett, C.S., Van Dongen, J .L., and
Thompson, C.B. (1996). CD30 contains two binding sites with differ-
ent specificities for members of the tumornecrosis factorreceptor-
associated factor family of signal transducing proteins. J . Biol.
Chem. 271, 12852±12858.
Hsu, H., Xiong, J ., and Goeddel, D.V. (1995). The TNF receptor
1-associated protein TRADD signals cell death and NF-?B activa-
tion. Cell 81, 495±504.
Hsu, H., Shu, H.-B., Pan, M.-G., and Goeddel, D.V. (1996a). TRADD-
TRAF2 and TRADD-FADD interactions define two distinct TNF re-
ceptor 1 signal transduction pathways. Cell 84, 299±308.
Hsu, H., Shu, H.-B., Baichwal, V., and Goeddel, D.V. (1996b). TNF-
dependent recruitment of the protein kinase RIP to the TNF
receptor-1 signaling complex. Immunity 4, 387±396.
Hsu, H., Solovyev, I.,Colombero, A., Elliott, R., Kelley, M., and Boyle,
W.J . (1997). ATAR, a novel tumor necrosis factor receptor family
member, signals through TRAF2 and TRAF5. J . Biol. Chem. 272,
Hu, H.M., O'Rourke, K., Boguski, M.S., and Dixit, V.M. (1994). A
novel RING fingerprotein interacts with the cytoplasmic domain of
CD40. J . Biol. Chem. 269, 30069±30072.
Ishida, T.K., Tojo, T.,Aoki, T.,Kobayashi, N.,Ohishi, T., Watanabe,T.,
Yamamoto, T., and Inoue, J . (1996a). TRAF5, a novel tumornecrosis
factorreceptor-associatedfactorfamily protein, mediates CD40 sig-
naling. Proc. Natl. Acad. Sci. USA 93, 9437±9442.
Ishida, T., Mizushima, S., Azuma, S., Kobayashi, N., Tojo, T., Suzuki,
K., Aizawa, S., Watanabe, T., Mosialos, G., Kieff, E., et al. (1996b).
Identification of TRAF6, a novel tumor necrosis factor receptor-
associated factor protein that mediates signaling from an amino-
terminal domain of the CD40 cytoplasmic region.J .Biol. Chem. 271,
Izumi, K.M., Kaye, K.M., and Kieff, E.D. (1997). The Epstein-Barr
virus LMP1aminoacid sequencethat engages tumornecrosisfactor
receptor associated factors is critical for primary B lymphocyte
growth transformation. Proc. Natl. Acad. Sci. USA 94, 1447±1452.
Karasuyama, H., and Melchers, F. (1988). Establishment of mouse
cell lines which constitutively secrete large quantities of interleu-
kin-2, 3, or5 using modified cDNA expressionvectors. Eur.J . Immu-
nol. 18, 97±104.
Kawabe, T., Naka, T., Yoshida, K., Tanaka, T., Fujiwara, H.,
Suematsu, S., Yoshida, N., Kishimoto, T., and Kikutani, H. (1994).
The immune responses in CD40-deficient mice: impaired immuno-
globulin class switching and germinal center formation. Immunity
Kyriakis, J .M., Banerjee, P., Nikolakaki, E., Dai, T., Rubie, E.A., Ah-
mad, M.F., Avruch, J ., and Woodgett, J .R. (1994). The stress-acti-
vated protein kinase subfamily of c-J un kinases. Nature 369,
Lee, S.Y., Lee, S.Y., Kandala, G., Liou, M.L., Liou, H.C., and Choi,
Y. (1996). CD30/TNF receptor-associated factor interaction: NF-
kappa B activation and binding specificity. Proc. Natl. Acad. Sci.
USA 93, 9699±9703.
Liu, Z.-G., Hsu, H., Goeddel, D.V., and Karin, M. (1996). Dissection
of TNF receptor 1 effector functions: J NK activation is not linked
to apoptosis while NF-?B activation prevents cell death. Cell 87,
Marsters, S.A., Ayres, T.M., Skubatch, M., Gray, C.L., Rothe, M.,
and Ashkenazi, A. (1997). Herpesvirus entry mediator, a member
fo the tumor necrosis factor receptor (TNFR) family, interacts with
members of the TNFR-associated factor family and activates the
transcription factors NF-?B and AP-1. J . Biol. Chem. 272, 14029±
Montgomery, R.I., Warner, M.S., Lum, B.J ., and Spear, P.G. (1996).
Herpes simplex virus-1 entry into cells mediated by a novel member
of the TNF/NGF receptor family. Cell 87, 426±436.
Mosialos,G., Birkenbach, M.,Yalamanchili, R., VanArsdale,T., Ware,
C., and Kieff, E. (1995). The Epstein-Barr virus transforming protein
LMP1 engages signaling proteins for the tumor necrosis factor re-
ceptor family. Cell 80, 389±399.
Nakano, H., Oshima, H., Chung, W., Williams-Abbott, L., Ware, C.F.,
Yagita, H., and Okumura, K. (1996). TRAF5, an activator of
NF-kappaB and putative signal transducerfor the lymphotoxin-beta
receptor. J . Biol. Chem. 271, 14661±14664.
Natoli, G., Costanzo, A., Ianni, A., Templeton, D.J ., Woodgett, J .R.,
Balsano, C., and Levrero, M. (1997). Activation of SAPK/J NK by
TNF receptor1 through a noncytotoxic TRAF2-dependentpathway.
Science 275, 200±203.
Nishina, H., Fischer, K.D., Radvanyi, L., Shahinian, A., Hakem, R.,
Rubie, E.A.,Bernstein, A.,Mak, T.W., Woodgett,J .R., and Penninger,
J .M.(1997). Stress-signalling kinase Sek1 protects thymocytes from
apoptosis mediated by CD95 and CD3. Nature 385, 350±353.
Pfeffer, K., Matsuyama, T., Kundig, T.M., Wakeham, A., Kishihara,
K., Shahinian, A., Wiegmann, K., Ohashi, P.S., Kronke, M., and Mak,
T.W. (1993).Mice deficientforthe 55 kdtumornecrosis factorrecep-
tor are resistant to endotoxic shock, yet succumb to L. monocyto-
genes infection. Cell 73, 457±467.
Probert, L., Keffer, J ., Corbella, P., Cazlaris, H., Patsavoudi, E., Ste-
phens, S., Kaslaris, E., Kioussis, D., and Kollias, G. (1993). Wasting,
ischemia, and lymphoid abnormalities in mice expressing T cell-
targeted human tumornecrosis factortransgenes. J .Immunol. 151,
Regnier, C.H., Tomasetto, C., Moog-Lutz, C., Chenard, M.P., Wen-
dling, C., Basset, P., and Rio, M.C. (1995). Presence of a new con-
served domain in CART1, a novel member of the tumor necrosis
factor receptor-associated protein family, which is expressed in
breast carcinoma. J . Biol. Chem. 270, 25715±25721.
Reinhard, C., Shamoon, B., Shyamala V., and Williams, L.T. (1997).
Tumor necrosis factor ?-induced activation of c-jun N-terminal ki-
nase is mediated by TRAF2. EMBO J . 16, 1080±1092.
Rothe, M., Wong, S.C., Henzel, W.J ., and Goeddel, D.V. (1994). A
novelfamily ofputative signal transducers associated with the cyto-
plasmic domain of the 75 kDa tumor necrosis factor receptor. Cell
Rothe, M., Sarma, V., Dixit, V.M., and Goeddel, D.V. (1995a). TRAF2-
mediated activation of NF-kappa B by TNF receptor 2 and CD40.
Science 269, 1424±1427.
Rothe, M., Pan, M.G., Henzel, W.J ., Ayres, T.M., and Goeddel, D.V.
(1995b). The TNFR2-TRAF signaling complex contains two novel
proteins related to baculoviral inhibitor of apoptosis proteins. Cell
Sato, T., Irie, S., and Reed, J .C. (1995). A novel memberof the TRAF
family of putative signal transducing proteins binds to the cytosolic
domain of CD40. FEBS Lett. 358, 113±118.
Schmid, I., Uittenbogaart, C.H., and Giorgi, J .V. (1994). Sensitive
method for measuring apoptosis and cell surface phenotype in hu-
man thymocytes by flow cytometry. Cytometry 15, 12±20.
Shu,H.-B., Takeuchi, M.,and Goeddel, D.V. (1996).The tumornecro-
sis factorreceptor 2 signal transducers TRAF2and c-IAP1 are com-
ponents of the tumor necrosis factor receptor 1 signaling complex.
Proc. Natl. Acad. Sci. USA 93, 13973±13978.
Smith, C.A., Farrah, T., and Goodwin, R.G. (1994). The TNF receptor
superfamily of cellular and viral proteins: activation, costimulation,
and death. Cell 76, 959±962.
Stanger, B.Z., Leder, P., Lee, T.H., Kim, E., and Seed, B. (1995). RIP:
a novel protein containing a death domain that interacts with Fas/
APO-1 (CD95) in yeast and causes cell death. Cell 81, 513±523.
Takeuchi, M., Rothe, M., and Goeddel, D.V. (1996). Anatomy of
TRAF2. Distinct domains for nuclear factor-kappaB activation and
association with tumor necrosis factor signaling proteins. J . Biol.
Chem. 271, 19935±19942.
Tartaglia, L.A., and Goeddel, D.V. (1992). Two TNF receptors. Immu-
nol. Today 13, 151±153.
Ting, A.T., Pimentel-Muinos, F.X., and Seed, B. (1996). RIP mediates
tumor necrosis factor receptor 1 activation of NF-?B but not Fas/
APO-1-initiated apoptosis. EMBO J . 15, 6189±6196.
Tsitsikov, E.N., Wright, D.A., and Geha, R.S. (1997). CD30 induction
of human immunodeficiency virus gene transcription is mediated
by TRAF2. Proc. Natl. Acad. Sci. USA 94, 1390±1395.
Van Antwerp, D.J ., Martin, S.J ., Kafri, T., Green, D.R., and Verma,
TNF-Induced Cell Death in TRAF2-Deficient Mice
I.M. (1996). Suppression of TNF-?-induced apoptosis by NF-?B.
Science 274, 787±789.
Vandenabeele, P., Declercq, W., Beyaert, R., and Fiers, W. (1995).
Two tumornecrosis factorreceptors: structure and function. Trends
Cell Biol. 5, 392±399.
Wang, C.-Y., Mayo, M.W., and Baldwin, A.S., J r. (1996). TNF- and
cancer therapy-induced apoptosis: potentiation by inhibition of
NF-?B. Science 274, 784±787.
Xu, Y., Cheng, G., and Baltimore, D. (1996). Targeted disruption
of TRAF3 leads to postnatal lethality and defective T-dependent
immune responses. Immunity 5, 407±415.