Cell, Vol. 88, 355–365, February 7, 1997, Copyright ã1997 by Cell Press
Apoptosis by Death Factor Review
Shigekazu Nagata family, while the length and sequence of the cytoplasmic
Department of Genetics segments differ significantly.
Osaka University Medical School Proteolysis of membrane-associated TNF produces
2-2 Yamada-oka, Suita soluble TNF. Theproteolysis is mediated by a membrane
Osaka 565 metalloproteinase (Gearing et al., 1994). Similarly, memJapan
brane-bound FasL undergoes metalloproteinase-mediOsaka
Bioscience Institute ated proteolytic cleavage to generate soluble cytokine
6-2-4 Furuedai, Suita (Tanaka et al., 1996). Aspecific metalloproteinase inhibiOsaka
565 tor blocks the processing of TNF as well as FasL, sugJapan
gesting that a similar enzyme cleaves TNF and FasL.
Since the CD40 ligand is also cleaved off from the membrane
to become soluble, it is likely that all TNF family
Introduction members are processed to a soluble form. The soluble
form of human FasL is functional, but mouse FasL loses
There is an old Japanese saying that “Once we are in its activity when it is cleaved from the membrane. Furthe
land of the living, we will eventually die.” This is true, thermore, membrane-bound TNF is more active than
not only for human beings, but also for the cells that soluble TNF in activating the type II TNF receptor (Grell
constitute our bodies. By repeated cell division (mitosis) et al., 1995). These results may indicate that FasL and
and differentiation, a fertilized egg produces billions of TNF work locally via cell–cell interactions under physiocells
to create our bodies. During this process, many logical conditions and that the purpose of shedding TNF
surplus or harmful cells are generated, and they must or FasL is to attenuate the process.
be removed or killed (Jacobson et al., 1997 [this issue The functional, soluble forms of TNFs as well as huof
Cell]. For example, thymocytes that have failed to man FasL exist as trimers (Tanaka et al., 1997). It has
rearrange their T cell–receptor gene, or whose T cell not yet been demonstrated whether membrane-bound
receptor may recognize their own tissues, must be elimi- TNF or FasL are trimers. However, lymphotoxin b, a
nated. The magnitude of the cell death is staggering: member of the TNF family, consists of a heterotrimer
of one a (lymphotoxin-a, or TNFb) and two b chains
more than 95% of thymocytes die in the thymus during
(lymphotoxin-b) on the membrane (Androlewicz et al.,
maturation. Even in adults, senescent cells are removed
1992), suggesting that membrane-bound TNF and FasL
and replaced by newly generated cells to maintain hohave
the potential to form trimeric structures. X-ray
meostasis. The cell death that occurs during emdiffraction
analyses of TNFa and TNFb have indicated
bryogenesis, metamorphosis, endocrine-dependent tisthat
each monomer forms an elongated, antiparallel
sue atrophy, and normal tissue turnover is “programmed
b-pleated sheet sandwich with a jelly roll topology
cell death,” mediated by a process termed “apoptosis.”
(Jones et al., 1989). Amino acids conserved among
Here, I focus on apoptosis controlled by cytokines.
members of the TNF family are mainly within the b Two death factors, Fas ligand (FasL) or tumor necrosis
strands. Computer-assisted modeling of FasL based on
factor (TNF), bind to their receptors and induce
the amino acid sequence suggests that FasL has a simiapoptosis,
killing the cells within hours. In a classic defi- lar tertiary structure to TNFa and TNFb.
nition of apoptosis, cells die by “suicide;” that is, cells The Fas and TNF Receptor Family
programmed to die would do so autonomously. How- Fas (also known as APO-1 or CD95), the receptor for
ever, the identification of death factor–receptor pairs FasL, is a type I–membrane protein (Itoh et al., 1991;
that regulate apoptosis indicates that apoptosis can Oehm et al., 1992) and a member of the TNF receptor
also becontrolled by an external killer in someinstances. (TNFR) family, which includes two TNFRs (TNFR1 and
TNFR2), the receptor for lymphotoxin-b, the NGF recep-
Death Factor and Receptor tor (p75), CD40, CD27, and CD30 (Nagata and Golstein,
Fas Ligand and the TNF Family 1995). This family is still growing, and three new mem-
Cytokines are a family of proteins that regulate cellular bers have recently been identified. They are human DR-3
proliferation and differentiation by binding to their spe- (death receptor-3)/Wsl-1 (Chinnaiyan et al., 1996; Kiston
cific receptors on target cells. Cytokines are grouped et al., 1996), human HVEM (herpes virus early mediator)
into at least three subfamilies based on structure, cyste- (Montogomery et al., 1996), and chicken CAR1 (cytoine-
knot growth factors, tumor necrosis factor, and heli- pathic avian leukosis-sarcoma virus receptor) (Brojatsch
cal cytokines. FasL belongs to the TNF family (Suda et et al., 1996). The extracellular region of the TNF receptor
al., 1993; Nagata and Golstein, 1995), which includes family members carries 2–6 repeats of a cysteine-rich
TNF, lymphotoxin, CD30 ligand, 4-1BB ligand, CD40 subdomain that has about25% similarity among various
ligand, CD27 ligand, and TRAIL (TNF-related apoptosis- members. In contrast, the cytoplasmic regions have little
inducing ligand). FasL is synthesized as a type II–mem- similarity among the members, except for Fas, TNFR1,
brane protein; that is, its N terminus is in the cytoplasm DR-3/Wsl-1, and CAR1, as discussed below.
and its C-terminal region extends into the extracellular TNF induces apoptosis and activates the transcription
space. The extracellular region of about 150 amino acids factor NF-kB. It can also stimulate the proliferation of
is well conserved (20–25%) among members of the TNF thymocytes. Although both TNFR1 and TNFR2 can
Cell
356
transduce the signal for apoptosis and NF-kBactivation, Fas- and TNFR1-mediated apoptosis occur in the
TNFR1 is responsible for these signals in most cases presence of inhibitors of either RNA or protein synthesis
(Vandenabeele et al., 1995). On the other hand, TNFR2 (Yonehara et al., 1989; Itoh et al., 1991). Even enucleated
but not TNFR1 is responsible for the TNF-induced prolif- cells undergo apoptosis upon Fas activation (Schulzeeration
signal in thymocytes. Binding of FasL to Fas or Osthoff et al., 1994), suggesting that all of the compocross-
linking Fas with agonistic antibodies (IgM class nents necessary for apoptotic signal transduction are
anti-Fas antibody, or IgG3 class anti-APO1 antibody) present and that Fas activation simply triggers this mainduces
apoptosis in Fas-bearing cells (Trauth et al., chinery. To dissect the signal-transducing machinery for
1989; Yonehara et al., 1989; Itoh et al., 1991). Most other Fas- and TNFR1-mediated apoptosis, two approaches
receptors in the TNF receptor family transduce activa- have been used. In one approach, several groups have
tion or stimulatory signals, although some of them, such identified a molecule(s) that binds to the cytoplasmic
as CD40 and CD30, may also have the ability to inhibit region of Fas or TNFR1, while in the other, information
growth, probably causing apoptosis. The presence of gained from studying apoptosis in the nematode, C.
a homologous domain (about 80 amino acids) in the elegans, was applied to the Fas and TNF system.
cytoplasmic regions of Fas and TNFR1 suggested that Utilization of the yeast two-hybrid system with the Fas
this region is responsible for transducing the death sig- cytoplasmic region as bait led to the identification of
nal. In fact, subsequent mutational analyses in Fas and a molecule called FADD (Fas-associating protein with
TNFR1 indicated that this is the case, and this domain death domain) or MORT1, which contains a death dohas
been designated a death domain (Itoh and Nagata, main at its C terminus (Boldin et al., 1995; Chinnaiyan
1993; Tartaglia et al., 1993). DR-3/Wsl-1 also carries a et al., 1995). FADD/MORT1 is recruited to Fas upon its
death domain and has the potential to transduce an activation (Kischkel et al., 1995) and binds to Fas via
apoptotic signal, as well as to activate NF-kB (Chinnai- interactions between the death domains. The N-terminal
yan et al., 1996; Kiston et al., 1996). CAR1, which also region (termed the death effector domain [DED] or
contains a death domain, has been shown to cause MORT1 domain) is responsible for downstream signal
apoptosis in chicken cells when it is cross-linked by the transduction. A similar death domain–containing protein
envelope protein of ALSV (Brojatsch et al., 1996). (TRADD, TNFR1-associated death domain protein)
The death domain has a tendency to self-aggregate,
binds to TNFR1 (Hsu et al., 1995). But unlike FADD/
and the tertiary structure of the Fas death domain, re-
MORT1, TRADD does not carry a death effector domain,
vealed by heteronuclear multidimensional NMR specand
its death domain is responsible for mediating
troscopy, shows that the death domain is a novel protein
apoptosis. This apparent discrepancy between FADD/
fold consisting of six antiparallel, amphipathic a helices
MORT1 and TRADD is resolved by the finding that
(Huang et al., 1996). Many charged amino acids are
TRADD binds to FADD/MORT1 present on the surface, which is probably responsible via interactions between
for mediating the interactions between death domains their death domains (Hsu et al., 1996b). These results
described below. suggest that Fas and TNFR1 use FADD as a common
signal transducer and share the signaling machinery
Signal for Apoptosis downstream of FADD/MORT1 (Figure 1). In addition to
Cascade Leading to ICE this pathway, TNFR1 has another pathway leading to
Binding of ligand to a tyrosine kinase receptor, such apoptosis. RIP (receptor interacting protein), originally
as PDGF or EGF receptor, induces dimerization of the identified as a Fas-binding protein, preferentially binds
receptor and activates the intrinsic kinase activity in the to TRADD (Hsu et al., 1996a). RIP is a serine/threonine
cytoplasmic domain. The receptors for hematopoietic kinase containing a death domain and binds to TRADD
growth factors such as colony-stimulating factor and via interactions between their death domains. RIP infor
interferons do not contain kinase domains in their duces apoptosis when overexpressed. The death docytoplasmic
regions. Instead, the ligand-induced dimer- main of RIP, but not its kinase domain, is responsible
ization recruits a kinase(s) to the receptor and activates for transduction of the death signal, indicating that RIP
it, which then results in transduction of the proliferation does not possess a death effector domain, but rather
and/or differentiation signals. In the case of Fas or another downstream effector molecule may be recruited
TNFR1, however, dimerization with a divalent anti-Fas or through the death domain of RIP (see below) (Figure 1).
TNFR1 monoclonal antibody is not sufficient to activate To find the signaling molecule downstream of FADD/
these receptors. Fas and TNFR1 must be oligomerized MORT1, Wallach and his associates again used the
to be activated; that is, IgM class anti-Fas monoclonal yeast two-hybrid system, using the N-terminal DED/
antibody or IgG3 class anti-APO1 antibody that possess MORT1 domain of FADD/MORT1 as bait (Boldin et al.,
a tendency to aggregate function as potent agonists 1996). At the same time, a collaborative group, led by
(Trauth et al., 1989; Yonehara et al., 1989). X-ray diffrac- Dixit and Peter, continued the biochemical characterization
analysis of the TNFb–TNF receptor complex has tion of molecules recruited to the activated Fas receptor
indicated that a TNFb trimer makes a complex with three (Muzio et al., 1996). Both groups identified the same
molecules of the extracellular region of the TNF receptor molecule, which was originally termed FLICE (FADD(
Banner et al., 1993), suggesting that TNF induces tri- like ICE) or MACH (MORT1-associated CED-3 homomerization
of the receptor. The similarity between the logue) and is now designated caspase-8 (Alnemri et
structures of FasL and TNF and between Fas and the al., 1996) (Table 1). Caspase-8 carries two DED/MORT1
TNF receptors suggests that FasL also induces trimeri- domains at the N-terminal region, through which it binds
zation of Fas and that the trimerized cytoplasmic region FADD/MORT1. The C-terminal region of caspase-8 is
then transduces the signal. related to ICE family members, more specifically, to
Review: Apoptosis by Death Factor
357
Figure 1. Models for Apoptosis Signaling by Death Factors
(A) Fas-induced apoptosis. Binding of FasL to Fas induces trimerization of the Fas receptor, which recruits caspase-8 (FLICE/MACH) via an
adaptor, FADD/MORT1. The oligomerization of FLICE may result in self-activation of proteolytic activity and trigger the ICE protease cascade.
The activated ICE members can cleave various substrates, such as poly(ADP) ribose polymerase (PARP), lamin, rho-GDI, and actin, and cause
morphological changes to the cells and nuclei.
(B) TNF-induced apoptosis. TNF binds to TNFR1, and the trimerized receptor recruits TRADD via interactions between death domains. The
death domain of TRADD then recruits FADD/MORT1 in one pathway to activate caspase-8. In another pathway, RIP binds to TRADD and
transduces an apoptotic signal through the death domain. In addition, RIP together with TRAF2 activates NF-kB, which may induce the
expression of survival genes. The role of the kinase activity of RIP is currently unknown.
members of the caspase-3 (CPP32) subfamily, and re- self-activation of the protease domain. One apoptotic
pathway from TNFR1 uses combinant caspase-8 preferentially cleaves caspase-3 caspase-8 pathway through
substrates over caspase-1 (ICE) substrates (Boldin et the interaction of TRADD with FADD/MORT1. TRADD
al., 1996). additionally recruits RIP, which may trigger a second
Figure 1 presents the current model for Fas- and apoptotic pathway. The recently identified DR-3/Wsl-1
TNFR1-mediated apoptosis. Binding of a trimeric FasL receptor is more similar to TNFR1 than to Fas. That is,
to Fas induces trimerization of Fas, and FADD/MORT1 DR-3 binds TRADD, which then recruits FADD and RIP
binds to the trimerized Fas cytoplasmic region through (Chinnaiyan et al., 1996; Kiston et al., 1996). The apopthe
interaction of the respective death domains. Cas- totic signaling pathway downstream of RIP is currently
pase-8 is then recruited to FADD/MORT1 through bind- unknown. However, another death domain–containing
ing of the DED domains, which in turn may induce adaptor, termed RAIDD (RIP-associated Ich-1/CED-3
Table 1. Human ICE Protease Superfamily
Recognition
Proteases Alternative Names Sequence Substrates
caspase-1 ICE YVAD pro-IL1b, pro-caspase 3 and 4
caspase-4 ICErel-II, TX, ICH-2
caspase-5 ICErel-III, TY
caspase-2 ICH-1 PARP
caspase-9 ICE-LAP6 PARP
caspase-3 CPP32, Yama, apopain DEVD PARP, DNA-PK, SRE/BP, rho-GDI
caspase-6 Mch2 VEID lamin A
caspase-7 Mch3, ICE-LAP3, CMH-1 PARP, pro-caspase 6
caspase-8 FLICE, MACH, Mch5
caspase-9 ICE-LAP6, Mch6 PARP
caspase-10 Mch4
The caspase family members can be divided into three subfamilies: caspase-1 (ICE), caspase-2 (ICH-1), and caspase-3 (CPP32), according
to Alnemiri et al. (1996).
Cell
358
homologous protein with a death domain) has recently designated as caspases (cysteine aspases) (Table 1)
(Alnemri et al., 1996). So far, been identified (Duan and Dixit, 1997). RAIDD binds RIP recognition sequences for
through its death domain and recruits caspase-2 (Ich-1) three ICE family members have been identified. That is,
to RIP. Although an involvment of RAIDD in the TNFR1 caspase-1 (ICE) recognizes the sequence Tyr–Val–Ala–
or DR3/Wsl-1-mediated apoptotic pathway has not yet Asp (YVAD) in the proform of IL-1b, caspase-3 (CPP32/
been demonstrated, it is possible that RAIDD plays a Yama/apopain) recognizes Asp–Glu–Val-Asp (DEVD)
role in transducing an apoptotic signal from one of the and cleaves poly(ADP-ribose) polymerase, and casdeath
receptors. pase-6 (Mch2) recognizes Val–Glu–Ile–Asp (VEID) and
The signal from Fas seems to be restricted to cleaves lamin (Nicholson et al., 1995; Takahashi et al.,
apoptosis, whereas other members of the TNF receptor 1996). However, it is uncertain whether each ICE family
family including TNFR1 activate NF-kB. NF-kB activa- member has a specific substrate for mediating apoption
by TNF receptor family members is mediated by tosis, or if some members of the subfamily are redun-
TRAF (TNF receptor–associated factor) family (Rothe et dant, cleaving the same substrates. In this regard, it is
al., 1994). So far, five members have been identified in noteworthy that caspase-1-null mice do not show any
this family, and all contain a TRAF domain of about 230 phenotype in programmed cell death (Li et al., 1995),
amino acids. Among members of this family, TRAF2 while the mice lacking caspase-3 show hyperplasia and
binds directly to TNFR2 and CD30 and indirectly to disorganized cell development in the brain (Kuida et al.,
TNFR1 through TRADD and RIP. A dominant-negative 1996). These results suggest that caspase-1 is redun-
TRAF2 blocks TNF-induced NF-kB activation, but not dant in all cell types, while caspase-3 plays a major role
apoptosis (Liu et al., 1996). Instead, blocking NF-kB in apoptosis in some cells of the brain.
activation with the dominant-negative TRAF2 potenti- Using what was known about the specific recognition
ates the cytotoxic activity of TNF in various cell types, sequences of the ICE proteases, specific competitive
suggesting that NF-kB activation leads to the expres- inhibitors and fluorescent substrates for caspase-1
sion of a protein(s) that inhibits TNF-induced cytotoxic- and -3 have been designed (Thornberry et al., 1992). In
ity. NF-kB consists of two subunits (p50 and p65) and addition, several proteins encoded by viral genes are
exists in a complex with IkB in resting cells. The signal known to inhibit members of the ICE family. These infrom
TRAF2 results in phosphorylation of IkB and subse- clude crmA, a cytokine response–modifier gene enquent
degradation by the proteosome. NF-kB, thus re- coded by cowpox virus, and p35, coded for by Baculovileased
from IkB, enters the nucleus and activates vari- rus. These viral proteins seem to inhibit protease activity
ous genes carrying the NF-kB response element. Cells by forming a stable complex. p35 has a broader specificlacking
one component of NF-kB (65 kDa) or expressing ity for ICE family members than crmA. That is, crmA
IkB mutants that cannot be phosphorylated are more preferentially inhibits caspase-1 over caspase-3, while
sensitive to TNF-induced cytotoxicity, confirming that p35 inhibits both caspase-1 and -3 equally well.
one of the target genes for NF-kB is a gene encoding Inhibitors of caspase-1 or -3 block Fas- and TNFa
survival factor (Beg and Baltimore, 1996; Liu et al., induced apoptosis, which suggests that both cas1996;
Van Antwerp et al., 1996; Wang et al., 1996). These pase-1- and caspase-3-like proteases are involved in
results are in good agreement with the fact that Fas, Fas- and TNFR1- mediated apoptosis (Enari et al.,
which cannot activate NF-kB,mediates a stronger apop- 1995b; Los et al., 1995; Tewari and Dixit, 1995; Enari et
totic signal than TNFR1, which can activate NF-kB. The al., 1996). Monitoring the protease activity with specific
cytotoxicity of TNF can be potentiated by cycloheximide fluorescent substrates for caspase-1 and -3 demonor
actinomycin D, which is probably due to the inhibition strates that a caspase-1-like protease is transiently actiof
the NF-kB-induced gene expression. vated, whereas the activation of a caspase-3-like prote-
ICE Protease Cascade ase gradually increases during Fas-induced apoptosis
Genetic analysis of programmed cell death in C. elegans (Enari et al., 1996). A similar sequential activation of
has revealed a number of gene products that regulate caspase-1- and caspase-3-like proteases was also
the cell death process (Ellis et al., 1991). Among them, found in vivo. When agonistic anti-Fas antibody was
the CED-3 product is required for cell death, and molec- administered to mice, the livers were damaged (Ogaular
cloning of the ced-3 gene revealed it to be a homo- sawara et al., 1993). As the damage proceeded, caslogue
of mammalian ICE (interleukin-1b converting en- pase-1-like activity was detected in the liver, followed
zyme) (Yuan et al., 1993), which converts the IL-1b by the gradual activation of a caspase-3-like protease
precursor to the mature form. ICE is a cysteine protease (Rodriguez et al., 1996a). The activation of the caspase-
consisting of two large (p17) and two small (p10) sub- 3-like protease is dependent on the activation of a casunits,
which are generated by proteolytic cleavage of pase-1-like protease (Enari et al., 1996), indicating that
the ICE precursor (a zymogen). Cross-hybridization with these proteases are sequentially activated. This sequen-
ICE cDNA and a search of the human genome database tial activation can also be seen in a cell-free system.
revealed at least 10 ICE homologues (see Table 1), which That is, cell lysate from Fas-activated, but not from
are divided into three subgroups (ICE-like, CPP32-like, nonactivated cells, induced apoptotic morphological
and Ich1-like proteases), based on their sequence ho- changes in intact nuclei (Enari et al., 1995a). However,
mology (Alnemri et al., 1996). All of these cause apop- when the cell lysates from growing, nonapoptotic cells
tosis when overexpressed in cells. They appear to be were supplemented with recombinant caspase-1 or -3,
cysteine proteases, containing conserved sequences the lysates induced apoptosis. This caspase-1-induced
for substrate binding and catalysis; they cleave their apoptosis was inhibited, not only by an inhibitor of cassubstrates
after aspartic acid. Therefore, they are now pase-1, but also by the inhibitor of caspase-3 (Enari et
Review: Apoptosis by Death Factor
359
al., 1996), confirming the sequential activation of cas- However, little is known about the biochemical mechapase-
1- and caspase-3-like proteases. It is likely that nism whereby CED-9/Bcl-2 and their family members
other members of the ICE family are also activated in inhibit apoptosis. Bcl-2 and Bcl-x are localized to outer
the cascade, cleaving their “death substrates” such as mitochondrial membranes and endoplasmic reticulum
lamin, actin, poly(ADP)ribose polymerase, rho-GDI, as well as nuclear membranes. The tertiary structure of
SREBP, and DNA-dependent protein kinase, to cause Bcl-xL has been determined by X-ray and NMRanalyses
the apoptotic morphological changes observed on cells (Muchmore et al., 1996). It consists of two central, hyand
nuclei, as well as chromosomal DNA degradation. drophobic a helices, which are similar to the pore-form-
As discussed above, Fas engagement recruits cas- ing bacteria toxins such as diphtheria toxin and the
pase-8 to the Fas receptor complex. How can this result colicins, suggesting that Bcl-xL also generates pores in
be integrated into the model of sequential ICE protease the membrane. When mitochondria are damaged by an
activation? Here, I suggest two models. In the first agent that causes permeability transition, nuclear apopmodel,
the oligomerization of caspase-8 through the tosis is induced (Zamzami et al., 1996). This permeability
interaction with FADD/MORT1 leads to its autocatalytic transition of mitochondrial membrane, and thus nuclear
activation, which then triggers the protease cascade by apoptosis, is blocked by Bcl-2, suggesting that the
cleaving the caspase-1-like protease zymogen. In the membrane pores in the mitochondria, generated by
second model, oligomerization does not activate cas- the Bcl-2 family members, play an important role in
pase-8, but a caspase-1-like protease activates the apoptosis, at least in this system.
oligomerized caspase-8, which then sequentially acti- Bcl-2 and Bcl-xL can also inhibit Fas-mediated apopvates
other members of the ICE family. In addition to tosis in vitro as well as in vivo (Itoh et al., 1993; Boise
ICE family proteases, other proteases such as cathepsin et al., 1995; Rodriguez et al., 1996b). Fas activation dam-
D aspartic protease and the serine protease AP24 ages mitochondrial function, but the damage is inhibited
(apoptosis protease 24) may be involved in Fas- and by ICE protease inhibitors (Krippner et al., 1996). These
TNFR1-induced apoptosis. To understand how these results suggest that the mitochondrial damage is downproteases
may be involved in the apoptotic process, stream of the ICE protease cascade in Fas-induced
it will be necessary to biochemically characterize the apoptosis and is probably a secondary effect. Thus, it
purified or recombinant proteins and determine their is not clear how Bcl-2/Bcl-xL located in mitochondria
specific substrates.
can modulate the Fas-induced apoptotic signaling path-
The Bcl-2 Family
way that seems to take place in the cytoplasm. One
ced-9, a homologue of the mammalian protooncogene
possible mechanism is that the damage of mitochondria
Bcl-2, prevents programmed cell death in C. elegans
by ICE protease may amplify the signal by releasing
(Hengartner and Horvitz, 1994). Similarly, overexpresapoptosis-
inducing molecules (Krippner et al., 1996;
sion of Bcl-2 blocks apoptosis of mammalian cells that
Zamzami is triggered by a number of different stimuli such as et al., 1996).
factor deprivation, irradiation, c-myc, or anti-cancer Other Regulators in the Signaling Pathway
drugs. A number of CED-9/Bcl-2 family members have Ceramide, generated by sphingomyelinases, increases
been identified in mammals: Bcl-2, Bcl-xL, Bcl-w, and during Fas- or TNFR1-mediated apoptosis, and ceraMcl-
1 inhibit apoptosis, whereas others, such as Bax, mide itself can induce cell death (Spiegel et al., 1996).
Bik, Bak, Bad, and Bcl-xs, activate apoptosis. The vari- Since ceramide activates the ras/MAP kinase pathway,
ous Bcl-2 family members can dimerize with one an- it was postulated that activated ras is responsible for
other, with one monomer antagonizing or enhancing the apoptotic cell death. However, the recent observation
function of the other. In this way, the ratio of inhibitors that generation of ceramideand activation of JNKduring
to activators in a cell may determine the propensity of Fas activation is blocked by ICE protease inhibitors sugthe
cell to undergo apoptosis (Yang and Korsmeyer, gests that the production of ceramide occurs down1996).
For example, if either bcl-x (Motoyama et al., stream of the ICE protease cascade (Gamen et al., 1996;
1995) or bcl-2 (Veis et al., 1993) is disrupted in mice, the Lenczowski et al., 1997). An increase in ceramide during
animals die as embryos or postnatally, respectively, as Fas activation is likely to be one of the changes that
the result of excessive programmed cell death in partic- accompanies apoptosis and is unlikely to be a mediator
ular organs. Conversely, if bax is disrupted, some normal of apoptosis. Many other proteins have been suggested
programmed cell death fails to occur (Knudson et al., as regulators of Fas-mediated apoptosis. For example,
1995). Another attractive mechanism to regulate dimer- c-abl tyrosine kinase, FAP tyrosine phosphatase, and
ization of Bcl-2 family members is phosphorylation (Ga- small stress proteins (HSP24) inhibit the process,
jewski and Thompson, 1996). For example, Bad, a pro- whereas the Fas-associated proteins of p59fyn kinase
apoptotic member of the Bcl-2 family, is phosphorylated and FAF seem to augment apoptotic signal induced by
by a putative kinase that can be activated by growth Fas. Howthese proteins regulate the process is currently
factor engagement. The phosphorylated Bad loses the unknown.
ability to bind Bcl-xL. Instead, it binds to 14-3-3, a protein
that can interact with several signaling enzymes.
The Bcl-xL dissociated from Bad now can execute its Physiological and Pathological Roles of Fas
antiapoptotic function (Zha et al., 1996). Down-Regulation of the Immune Reaction
How does Bcl-2 or Bcl-xL inhibit apoptosis? Genetic Apoptosis occurs in various processes in mammalian
studies of ced-9, ced-4, and ced-3 mutants in C. elegans life (Jacobson et al., 1997). What kinds of apoptosis
indicate that ced-9 controls programmed cell death up- are regulated by the Fas system? Fas is ubiquitously
stream of ced-4 and ced-3 (Shaham and Horvitz, 1996). expressed in various tissues with abundant expression
Cell
360
after the activated T cells accomplish their task, they
must be removed to avoid accumulation. Mature T cells
from lpr or gld mice do not die after activation, and
activated cells accumulate in the lymph nodes and
spleens of these mice. When T cell hybridomas are activated
in the presence of a Fas-neutralizing molecule,
they do not die. These results indicate that Fas is involved
in activation-induced suicide of T cells, i.e., in
down-regulation of the immune reaction (Figure 3A) (Nagata
and Golstein, 1995). Peripheral clonal deletion may
also be mediated by the Fas system, because the cells
to be deleted in this process are activated by interactions
with cells expressing self antigens. However, thymic
clonal deletion is apparently normal in mice lacking
the functional Fas system (lpr-, gld-, or Fas-null mice)
(Singer and Abbas, 1994), even though thymocytes
abundantly express Fas and are sensitive to Fasinduced
apoptosis. These results suggest that Fas is
not involved in the deletion process in the thymus, although
one cannot rule out the possibility that this process
is mediated by redundant mechanisms.
In addition to T cells, the Fas-deficient mice accumulate
B cells and have elevated levels of immunoglobulins
Figure 2. When Apoptosis Fails
of various classes that include anti-ssDNA and anti-
Wild-type (1/1) and Fas-null (2/2) mice were killed at 16 weeks of dsDNA antibodies (Cohen and age, and their lymph nodes (LN) and spleen (SP) are shown (from Eisenberg, 1991), sugAdachi
et al., 1995). gesting an involvement of the Fas system in the deletion
of activated or autoreactive B lymphocytes. In fact, immunization
of mice with antigens rapidly induces Fas
in the thymus, liver, heart, and kidney. On the other expression in germinal centers. Furthermore, the actihand,
FasL is predominantly expressed in activated T vation of naive B cells through CD40 sensitizes them
lymphocytes and Natural Killer (NK) cells, although it to Fas-mediated apoptosis, while their costimulation
is also expressed constitutively in the tissues of the through CD40 and Ig receptor makes them resistant
“immune-privilege sites” such as the testis and eye, as (Rothstein et al., 1995). Although these results suggest
described below. The mouse mutations, lpr (lympho- that FasL-expressing T cells kill the Fas-expressing actiproliferation)
and gld (generalized lymphoproliferative vated B cells, the precise mechanism and physiological
disease), are spontaneous recessive mutations (Cohen role of Fas in the deletion of B cells remains to be
and Eisenberg, 1991). Mice carrying homozygous muta- studied.
tions in lpr or gld develop lymphadenopathy and spleno- Children carrying a defect in the Fas gene have also
megaly by accumulating CD42CD82 cells of T cell origin, been identified (Fisher et al., 1995; Rieux-Laucat et al.,
and some strains of mice develop autoimmune dis- 1995). Most of these patients carry a heterozygous mueases.
Genetic and molecular analyses of lpr and gld tation in the Fas gene. The affected Fas protein seems
mutations showed that they are loss-of-function muta- to work in a dominant-negative fashion, and T cells from
tions in the Fas and FasL genes, respectively (Wata- the patients do not die upon activation. The patients
nabe-Fukunaga et al., 1992; Takahashi et al., 1994). The show phenotypes (ALPS, autoimmune lymphoprolifera-
Fas-null mice, established by gene targeting (Adachi et tive syndrome) that are remarkably similar to those of lpr
al., 1995), also show lymphadenopathy and splenomeg- mice, including lymphadenopathy, splenomegaly, and
aly (Figure 2), which is much more pronounced than in hypergammaglobulinemia. Some patients show autoimmice
carrying the leaky lpr mutation. Furthermore, when munediseases such as hemolytic anemia, thrombocyto-
Fas was expressed in the lymphocytes of lpr mice as a penia, and neutropenia by producing autoantibodies
transgene, the lymphoproliferative phenotype was res- against red blood cells and platelets. On the other hand,
cued (Wu et al., 1994), confirming that Fas plays a role the fathers or mothers of the patients, who also carry
in the programmed cell death of T lymphocytes. the heterozygous mutation of the Fas gene, do not show
T lymphocytes, which are responsible for removing an abnormal phenotype, suggesting that the patients
virally infected and cancerous cells, die at various carry mutations in other complementing genes. Alternastages
of their development. Most immature T cells are tively, Fas could berequired only for the perinatal period,
useless (incorrect rearrangement of the T cell receptor) and the parents may also have had a similar phenotype
or potentially detrimental (self-reactive) to the organism. in childhood that was rescued later, since the heterozy-
More than 95% of thymocytes that immigrate into the gous mutation is leaky.
thymus are eliminated by positive and negative selection Effector of Cytotoxic T Lymphocytes
during their development. In the periphery, mature T and Natural Killer Cells
cells that recognize self antigens are also deleted (pe- Cytotoxic T lymphocytes (CTL) recognize and kill cells
ripheral clonal deletion). When mature T cells encounter infected by viruses or bacteria, whileNK cells kill cancertarget
cells, they are activated to proliferate. However, ous cells. The professional CTL are CD81 T cells, but
Review: Apoptosis by Death Factor
361
Figure 3. Three Types of Killing by the Fas and FasL System
(A) Activation-induced suicide of T cells. Mature T lymphocytes are activated by T cell–receptor interaction with antigen-presenting cells. The
activated T cells express FasL, which binds to the Fas-expressing activated T cells to induce apoptosis.
(B) CTL-mediated killing of target cells. Virally infected cells present viral antigen as a complex with MHC. The cytotoxic T cells recognize
the antigen and become activated, leading to the expression of FasL. FasL then binds to Fas on the target cells to induce apoptosis.
(C) Killing of inflammatory cells in immune privilege sites and killing of CTL by tumor cells. Stromal cells in the immune privilege sites such
as the eye and testis and some tumor cells constitutively express FasL. When activated T cells or neutrophils enter an immune privileged
site, FasL binds to Fas on these cells and kills them to prevent inflammation. Similarly, when CTL or NK cells approach tumor cells, the tumor
cells counterattack these cells to escape from the immune destruction.
Th1-type CD41T cells also showcytotoxicity.How these effector cells binds to Fas on the target cell and causes
apoptosis by activating caspases, CTL and NK cells kill target cells was under debate for as described above
a long time, because a well-known perforin/granzyme- (Figure 3B). A similar activation of CTLs through T cell
based mechanism could not account for all of the exam- receptor would release perforins and granzymes that
ples of CTL killing. However, the identification of FasL were stored in granules. It is believed that perforin
as a cytotoxic molecule expressed by activated T cells makes pores in the plasmamembrane of the target cells,
resolved this problem. Studies with mice deficient in through which granzymes are introduced. One of the
either perforin/granzyme or FasL indicated that the per- granzymes (granzyme B) is a serine protease aspase,
forin/granzyme and FasL systems are major pathways which activates some of the caspase family members
for CTL-mediated cytotoxicity (Nagata and Golstein, by proteolysis (Darmon et al., 1995). Thus, although per-
1995). Activation of CTLs through T cell–receptor inter- forin/granzyme and FasL can independently trigger the
action with viral antigens induces the expression of the cell death program, the processes leading to apoptosis
FasL gene. The FasL expressed on the surface of the are similar in both cases. The CD81 T cells and NK cells
Cell
362
use both the perforin/granzyme and FasL/Fas pathways, Fas gene are not tumorigenic. However, the families of
these patients sometimes have whereas the Th1-type CD4 T cells preferentially use the histories of Hodgkin’s
FasL system. Whether particular CD81 T cells and NK lymphoma (Fisher et al., 1995). The abnormal survival
cells have any preference for using the perforin/gran- of the lymphocytes may allow the cells to accumulate
zyme or FasL/Fas system on specific target cells re- mutations that lead to malignancy. Genes for death facmains
to bestudied. Furthermore, CTLsin mice deficient tors and their receptors, such as FasL and Fas, may
in both the perforin and FasL systems show some resid- therefore be regarded as tumor suppressor genes. On
ual cytotoxicity in long-term assays (Braun et al., 1996), the other hand, when the system overfunctions, it
suggesting that yet another death factor(s), perhaps TNF causes tissue destruction and kills the animals. When
or TRAIL, functions as a CTL effector under these condi- an agonistic anti-Fas antibody or recombinant FasL was
tions. injected into mice to activate the Fas system in vivo, the
Immune Privilege mice were quickly killed by liver failure with symptoms
Cellular immune response reactions and their associ- similar to fulminant human hepatitis (Ogasawara et al.,
ated inflammatory responses can cause nonspecific 1993; Tanaka et al., 1997). Fulminant hepatitis is known
damage to nearby tissues. Although most organs can to be caused by abnormally activated T cells, and the
tolerate such inflammation, some, such as the eye and transformation of hepatocytes with hepatitis B virus or
testis, cannot. These organs, therefore, have a mecha- hepatitisCvirus causes the up-regulation of Fas expresnism
to protect themselves against dangerous and un- sion. These results suggest that under normal circumwanted
immune reactions. These organs are called “im- stances, CTLs recognizing viral antigens expressed on
mune privilege sites” and are able to support allogenic the cell surface of infected hepatocytes are activated
and xenogeneic tissue transplants. Initially, it was through the T cell receptor and kill the hepatocytes via
thought that immune privilege is maintained by pre- the Fas/FasL system. If this killing process works propventing
the activated cells from entering the organs. erly, it benefits the organism. However, when thesystem
However, another attractive mechanism has recently is exaggerated, it may lead to fulminant hepatitis. It is
been proposed (Bellgrau et al., 1995; Griffith et al., 1995). possible that other CTL-induced autoimmune diseases
That is, although the activated inflammatory cells can such as graft-versus-host disease, AIDS, and insulitis
enter these organs, they are immediately killed by FasL are also mediated by the Fas system.
expressed in the organs (Figure 3C). The constitutive Various cancer patients produce TNFa in a soluble
expression of functional FasL has been found in the form, and it works like a cachectin to induce systemic
corneal epithelium and endothelium, iris, and ciliary cells tissue damage. Similarly, the soluble form of FasL was
of the eye, as well as in the Sertoli cells of the testis. found in the sera of patients with NK lymphoma or large
When the eyes of wild-type mice were infected with granular lymphocytic leukemia (LGL) of the NK or T cell
herpes simplex virus (HSV-1), very few inflammatory type (Tanaka et al., 1996). The leukemic cells themselves
cells were found associated with the retina, while mas- were found to express functional FasL on their surfaces.
sive inflammation was observed in the retina of gld mice, These patients often show systemic tissue damage such
which have a defect in FasL. Furthermore, while testes as hepatitis and neutropenia. Since hepatocytes and
expressing functional FasL survived when transplanted neutrophils are particularly sensitive to Fas-mediated
under the kidney capsule of allogenic animals, testis apoptosis, it is possible that the systemic tissuedamage
grafts from gld mice were rejected. These findings indi- observed in these patients is due to FasL in their serum
cate that FasL accounts for at least part of the immune- or to FasL expressed on the circulating leukemic cells.
privileged nature of the eye and testis and suggest a
use for FasL as an immunosuppressive agent to target
activated effector cells in transplantation. In fact, when Conclusions and Perspectives
islets of Langerhans were cotransplanted with synge-
neic myoblasts expressing functional FasL, they were Many growth and differentiation factors regulate prolifprotected
from immune rejection and were able to main- eration and differentiation of mammalian cells during
tain normoglycemia for a substantial period in a mouse development. So far, three death factors (TNF, FasL,
model system for diabetes (Lau et al., 1996). Similarly, and TRAIL) and four death factor receptors (Fas, TNFR1,
several groups have recently found that some tumor DR3/Wsl-1, and CAR1) have been identified. Loss-ofcells
become resistant to Fas-induced apoptosis and function mutations in the Fas system, lpr and gld mice,
constitutively express FasL (Hahne et al., 1996; Strand illustrated the importance of this death factor system in
et al., 1996). FasL expressed on tumor cells then coun- maintaining mammalian homeostasis, specifically in the
terattacks CTL and NK cells by binding Fas on their life and death of lymphocytes. It is possible that many
surfaces to cause apoptosis. This mechanism may also more death factor and receptor systems that regulate
account for the ability of tumor cells to evade immune apoptosis in a tissue-specific manner will be found in
destruction. the future. Growth and differentiation signals are mediA
Double-Edged Sword ated by the phosphorylation and dephosphorylation of
As long as death factors are appropriately expressed, proteins, as well as by small second-messenger molethey
will be useful in maintaining homeostasis. However, cules such as cAMP and phosphatidyl inositol. These
if thesystem under- or over-functions, it will have delete- signals are reversible in most cases. On the other hand,
rious effects. Loss of function causes hyperplasia, such the apoptosis signal triggered by death factors is irreas
lymphoproliferation. The lymphocytes accumulated versible; that is, a protease cascade is activated by the
in patients carrying the heterozygous mutation in the death signal, and the proteases cleave various cellular
Review: Apoptosis by Death Factor
363
Japan and by a Research Grant from the components, which leads to morphological changes of Princess Takamatsu Canthe
cells and nuclei that are typical for apoptosis. Since cer Research Fund, and performed in part through Special Coordiother
apoptosis-inducing agents also activate caspase nation Funds of the Science and Technology Agency of the Japanese
Government. Because of limitations on the number of
family proteases, their signal transduction system may references, I cannot cite many important papers; I apologize to their
be similar or identical. However, the apoptotic system authors.
in mammals seems to be more complicated and sophis-
ticated than that in C. elegans. Instead of the single ICE/ References
CED-3 and the single Bcl-2/CED-9 in C. elegans, the
mammalian genome carries at least ten members of the Adachi, M., Suematsu, S., Kondo, T., Ogasawara, J., Tanaka, T.,
Yoshida, N., and Nagata, S. (1995). Targeted mutation in the Fas caspase family and nine members of the Bcl-2 family. gene causes hyperplasia in the peripheral lymphoid organs and
Whether they are just redundant or have different roles liver. Nature Genet. 11, 294–300.
remains to be examined. Biochemical analysis of each Alnemri, E.S., Livingston, D.J., Nicholson, D.W., Salvesen, G.,
family member and establishment of mice lacking each Thornberry, N.A., Wong, W.W., and Yuan, J. (1996). Human ICE/
member will clarify these points. Both TNF and FasL CED-3 protease nomenclature. Cell 87, 171.
induce apoptosis. However, since TNF can induce other Androlewicz, M.J., Browning, J.L., and Ware, C.F. (1992). Lymphosignals,
such as activation of NF-kB, it was thought likely toxin is expressed as a heteromeric complex with a distinct 33-kDa
that the signaling pathway through the TNF receptor glycoprotein on the surface of an activated human T cell hybridoma.
would be more complicated. In fact, identification of the J. Biol. Chem. 267, 2542–2547.
molecules involved in TNF and Fas signaling indicates Banner, D.W., D’Arcy, A., Janes, W., Gentz, R., Schoenfeld, H.-J.,
that the Fas-mediated signal is simpler than that of the Broger, C., Loetscher, H., and Lesslauer, W. (1993). Crystal structure
of the soluble human 55 kd TNF receptor–human TNFb complex:
TNF receptor. The TNF receptor shares a signal cascade implication for TNF receptor activation. Cell 73, 431–445.
with Fas in one apoptotic pathway, but it also activates Beg, A.A., and Baltimore, D. (1996). An essential role for NF-kB in
additional signaling pathways including one that acti- preventing TNF-a-induced cell death. Science 274, 782–784.
vates a survival signal. It will be interesting to examine Bellgrau, D., Gold, D., Selawry, H., Moore, J., Franzusoff, A., and
what kinds of survival genes are activated by NF-kBand Duke, R.C. (1995). A role for CD95 ligand inpreventing graft rejection.
how these molecules inhibit apoptosis. Identification of Nature 377, 630–632.
these survival genes may provide clues as to why some Boise, L.H., Minn, A.J., Noel, P.J., June, C.H., Accavitti, M.A., Lindtumor
cells are resistant to various apoptosis-inducing sten, T., and Thompson, C.B. (1995). CD28 costimulation can proagents
including FasL, TNF, and anti-cancer drugs. mote T cell survivalby enhancing the expression of Bcl-xL. Immunity
As described above, the Fas death factor system is 3, 87–98.
a double-edged sword. If this system is properly regu- Boldin, M.P., Varfolomeev, E.E., Pancer, Z., Mett, I.L., Camonis, J.H.,
lated, it is useful for down-regulating the immune reac- and Wallach, D. (1995). A novel protein that interacts with the death
domain of Fas/APO1 contains a sequence motif related to the death
tion and for removing virally infected as well as cancer- domain. J. Biol. Chem. 270, 7795–7798.
ous cells; but, if this system is exaggerated, it can cause Boldin, M.P., Goncharov, T.M., Goltsev, Y.V., and Wallach, D. (1996).
tissue destruction. How can modulation of this system Involvement of MACH, a novel MORT1/FADD-interacting protease,
be applied to human diseases? The first obvious appli- in Fas/APO-1- and TNF receptor–induced cell death. Cell 85,
cation is the killing of tumor cells, since some cancer 803–815.
cells, particularly some lymphoid tumors, express func- Braun, M.Y., Lowin, B., French, L., Acha-Orbea, H., and Tschopp,
tional Fas. However, since the systemic treatment of J. (1996). Cytotoxic T cells deficient in both functional Fas ligand
patients with FasL will cause deleterious side effects, and perforin show residual cytolytic activity yet lose their capacity
to induce lethal acute graft-versus-host disease. J. Exp. Med. 183, methods of local administration and/or proper targeting 657–661.
of FasL to the tumor should be devised. FasL can also
Brojatsch, J., Naughton, J., Rolls, M.M., Zingler, K., and Young, be used as an immune-suppressive agent. As discussed J.A.T. (1996). CAR1, a TNFR-related protein, is a cellular receptor for
above, the rejection of transplants is mediated by acti- cytopathic avian leukosis-sarcoma viruses and mediates apoptosis.
vated T cells. If a transplanted tissue is engineered to Cell 87, 845–855.
express FasL or is cotransplanted with FasL-expressing Chinnaiyan, A.M., O’Rourke, K., Tewari, M., and Dixit, V.M. (1995).
cells, the transplant may betolerated. The other applica- FADD, a novel death domain–containing protein, interacts w, ith the
tion of this system is to block FasL-induced tissue de- death domain of Fas and initiates apoptosis. Cell 81, 505–512.
struction. If Fas is shown to play a role in human diseases Chinnaiyan, A.M., O’Rourke, K., Yu, G.-L., Lyons, R.H., Garg, M.,
such as fulminant hepatitis, AIDS, and other diseases Duan, R.D., Xing, L., Gentz, R., Ni, J., and Dixit, V.M. (1996). Signal
involving CTL-induced tissue destruction, then neu- transduction by DR-3, a death domain–containing receptor related
to TNFR-1 and CD95. Science 274, 990–992. tralizing antibodies against Fas or FasL, or other inhibi-
Cohen, P.L., and Eisenberg, R.A. (1991). Lpr and gld: single gene tors of Fas-mediated apoptosis, would have potential models of systemic autoimmunity and lymphoproliferative disease.
as therapeutic agents. Annu. Rev. Immunol. 9, 243–269.
Darmon, A.J., Nicholson, D.W., and Bleackley, R.C. (1995). Activa-
Acknowledgments tion of the apoptotic protease CPP32 by cytotoxic T-cell-derived
granzyme B. Nature 377, 446–448.
I thank Drs. D. V. Goeddel and P. Golstein for helpful comments on
Duan, H., and Dixit, V.M. (1997). RAIDD, a novel death adaptor molethe
manuscript. I am grateful to all members of my laboratory in
cule. Nature 385, 86–89.
Osaka University Medical School and Osaka Bioscience Institute,
and to Dr. O. Hayaishi for encouragement and discussion. I thank Ellis, R.E., Yuan, J., and Horvitz, H.R. (1991). Mechanisms and funcDrs.
L. Migletta and G. Gray of Clarify Editing for careful editing of tions of cell death. Annu. Rev. Cell Biol. 7, 663–698.
the manuscript. The work in my laboratory was supported in part by Enari, M., Hase, A., and Nagata, S. (1995a). Apoptosis by a cytosolic
Grants-in-Aid from the Ministry of Education, Science, and Culture of extract from Fas-activated cells. EMBO J. 14, 5201–5208.
Cell
364
Enari, M., Hug, H., and Nagata, S. (1995b). Involvement of an ICE- Korsmeyer, S.J. (1995). Bax-deficient mice with lymphoid hyperplasia
and male germ cell death. Science like protease in Fas-mediated apoptosis. Nature 375, 78–81. 270, 96–99.
Enari, M., Talanian, R.V., Wong, W.W., and Nagata, S. (1996). Se- Krippner, A., Matsuno-Yagi, A., Gottlieb, R., and Babior, B. (1996).
quential activation of ICE-like and CPP32-like proteases during Fas- Loss of function of cytochrome c in Jurkat cells undergoing Fasmediated
apoptosis. Nature 380, 723–726. mediated apoptosis. J. Biol. Chem. 271, 21629–21636.
Fisher, G.H., Rosenberg, F.J., Straus, S.E., Dale, J.K., Middelton, Kuida, K., Zheng, T.S., Na, S.-Q., Kuan, C.-Y., Yang, D., Karasuyama,
L.A., Lin, A.Y., Strober, W., Lenardo, M.J., and Puck, J.M. (1995). H., Rakic, P., and Flavell, R.A. (1996). Decreased apoptosis in the
Dominant interfering Fas gene mutations impair apoptosis in a hu- brain and premature lethality in CPP32-deficient mice. Nature 384,
man autoimmune lymphoproliferative syndrome. Cell 81, 935–946. 368–372.
Gajewski, T.F., and Thompson, C.B. (1996). Apoptosis meets signal Lau, H.T., Yu, M., Fontana, A., and Stoeckert Jr., C.J. (1996). Preventransduction:
elimination of BAD influence. Cell 87, 589–592. tion of islet allograft rejection with engineered myoblasts expressing
Gamen, S., Marzo, I., Anel, A., Pin˜ eiro, A., and Naval, J. (1996). FasL in mice. Science 273, 109–112.
CPP32 inhibition prevents Fas-induced ceramide generation and Lenczowski, J.M., Dominguez, L., Eder, A., King, L.B., Zacharchuk,
apoptosis in human cells. FEBS Lett. 390, 233–237. C.M., and Ashwell, J.D. (1997). Lack of a role for Jun kinase and
AP-1 in Fas-induced apoptosis. Mol. Cell. Biol. 17, 170–181.
Gearing, A.J.H., Beckett, P., Christodoulou, M., Churchill, M., Clem-
ents, J., Davidson, A.H., Drummond, A.H., Galloway, W.A., Gilbert, Li, P., Allen, H., Banerjee, S., Franklin, S., Herzog, L., Johnston, C.,
R., Gordon, J.L., et al. (1994). Processing of tumor necrosis fac- McDowell, J., Paskind, M., Rodman, L., Salfeld, J., et al. (1995). Mice
tor-a precursor by metalloproteinases. Nature 370, 555–557. deficient in IL-1b-Converting enzyme are defective in production of
mature IL-1b and resistant to endotoxic shock. Cell 80, 401–411.
Grell, M., Douni, E., Wajant, H., Lo¨ hden, M., Maxeiner, B., Georgo-
poulos, S., Lesslauer, W., Kollias, G., Pfizenmaier, K., and Scheurich, Liu, Z.-G., Hsu, H., Goeddel, D., and Karin, M. (1996). Dissection of
P. (1995). The transmembrane form of the tumor necrosis factor is TNF receptor 1 effector functions: JNK activation is not linked to
the prime activating ligand of the 80 kDa tumor necrosis factor apoptosis while NF-kB activation prevents cell death. Cell 87,
receptor. Cell 83, 793–802. 565–576.
Griffith, T.S., Brunner, T., Fletcher, S.M., Green, D.R., and Ferguson, Los, M., Van de Craen, M., Penning, L.C., Schenk, H., Westendorp,
T.A. (1995). Fas ligand–induced apoptosis as a mechanism of im- M., Baeuerle, P.A.,Dro¨ ge, W., Krammer, P.H., Fiers, W., andSchulze-
mune privilege. Science 270, 1189–1192. Osthoff, K. (1995). Requirement of an ICE/CED-3 protease for Fas/
APO-1-mediated apoptosis. Nature 375, 81–83.
Hahne, M., Rimoldi, D., Schro¨ ter, M., Romero, P., Schreier, M.,
Montogomery, R.I., Warner, M.S., Lum, B.J., and Spear, P.G. (1996).
French, L.E., Schneider, P., Bornand, T., Fontana, A., Lienard, D.,
Herpes simplex virus-1 entry into cells mediated by a novel member
et al. (1996). Melanoma cell expression of Fas(Apo-1/CD95) ligand:
of the TNF/NGF receptor family. Cell 87, 427–436.
Implications for tumor immune escape. Science 274, 1363–1366.
Motoyama, N., Wang, F., Roth, K.A., Sawa, H., Nakayama, K., Naka-
Hengartner, M.O., and Horvitz, H.R. (1994). C. elegans cell survival yama, K., Negishi, I., Senju, S., Zhang, Q., Fujii, S., and Loh, D.Y.
gene ced-9 encodes a functional homolog of the mammalian proto- (1995). Massive cell death of immature hematopoietic cells and neuoncogene
bcl-2. Cell 76, 665–676. rons in Bcl-x-deficient mice. Science 267, 1506–1510.
Hsu, H., Xiong, J., and Goeddel, D.V. (1995). The TNF receptor Muchmore, S.W., Sattlet, M., Liang, H., Meadows, R.P., Harlan, J.E.,
1–associated protein TRADD signals cell death and NF-kB activa- Yoon, H.S., Nettesheim, D., Chang, B.S., Thompson, C.B., Wong,
tion. Cell 81, 495–504. S.L., Ng, S.L., and Fesik, S.W. (1996). X-ray and NMR structure of
Hsu, H., Huang, J., Shu, H.-B., Baichwal, V., and Goeddel, D. (1996a). human Bcl-xL, an inhibitor of programmed cell death. Nature 381,
TNF-dependent recruitment of the protein kinase RIP to the TNF 335–341.
receptor-1 signaling complex. Immunity 4, 387–396. Muzio, M., Chinnaiyan, A.M., Kischkel, F.C., O’Rourke, K., Shev-
Hsu, H., Shu, H.-B., Pan, M.-G., and Goeddel, D.V. (1996b). TRADD– chenko, A., Ni, J., Scaffidi, C., Bretz, J.D., Zhang, M., Gentz, R., et
TRAF2 and TRADD–FADD interactions define two distinct TNF re- al. (1996). FLICE, a novel FADD-homologous ICE/CED-3-like proteceptor
1 signal transcription pathways. Cell 84, 299–308. ase, is recruited to the CD95(Fas/APO-1) death-inducing signaling
Huang, B., Eberstadt, M., Olejniczak, E.T., Meadows, R.P., and complex. Cell 85, 817–827.
Fesik, S.W. (1996).NMR structure and mutagenesis of the Fas (Apo- Nagata, S., and Golstein, P. (1995). The Fas death factor. Science
1/CD95) death domain. Nature 384, 638–641. 267, 1449–1456.
Itoh, N., Yonehara, S., Ishii, A., Yonehara, M., Mizushima, S., Same- Nicholson, D.W., Ali, A., Thornberry, N.A., Vaillancourt, J.P., Ding,
shima, M., Hase, A., Seto, Y., and Nagata, S. (1991). The polypeptide C.K., Gallant, M., Gareau, Y., Griffin, P.R., Labelle, M., Lazebnik,
encoded by the cDNA for human cell surface antigen Fas can medi- Y.A., et al. (1995). Identification and inhibition of the ICE/CED-3
ate apoptosis. Cell 66, 233–243. protease necessary for mammalian apoptosis. Nature 376, 37–43.
Itoh, N., and Nagata, S. (1993). A novel protein domain required for Oehm, A., Behrmann, I., Falk, W., Pawlita, M., Maier, G., Klas, C.,
apoptosis: mutational analysis of human Fas antigen. J. Biol. Chem. Li-Weber, M., Richards, S., Dhein, J., Trauth, B.C., et al. (1992).
268, 10932–10937. Purification and molecular cloning of the APO-1 cell surface antigen,
Itoh, N., Tsujimoto, Y., and Nagata, S. (1993). Effect of bcl-2 on Fas a member of the tumor necrosis factor/nerve growth factor receptor
antigen–mediated cell death. J. Immunol. 151, 621–627. superfamily: sequence identity with the Fas antigen. J. Biol. Chem.
267, 10709–10715.
Jacobson, M.D., Weil, M., and Raff, M.C. (1997). Programmed cell
Ogasawara, J., Watanabe-Fukunaga, R., Adachi, M., Matsuzawa,
death in animal development. Cell, this issue.
A., Kasugai, T., Kitamura, Y., Itoh, N., Suda, T., and Nagata, S. (1993).
Jones, E.Y., Stuart, D.I., and Walker, N.P.C. (1989). Structure of Lethal effect of the anti-Fas antibody in mice. Nature 364, 806–809.
tumour necrosis factor. Nature 338, 225–228. Rieux-Laucat, F., Le Deist, F., Hivroz, C., Roberts, I.A., Debatin,
Kischkel, F.C., Hellbardt, S., Behrmann, I., Germer, M., Pawlita, M., K.M., Fischer, A., and de Villarty, J.P. (1995). Mutations in Fas associ-
Krammer, P.H., and Peter, M.E. (1995). Cytotoxicity-dependent ated with human lymphoproliferative syndrome and autoimmunity.
APO-1 (Fas/CD95)-associated proteins from a death-inducing sig- Science 268, 1347–1349.
naling complex (DISC) with the receptor. EMBO J. 14, 5579–5588. Rodriguez, I., Matsuura, K., Ody, C., Nagata, S., and Vassalli, P.
Kiston, J., Raven, T., Jiang, Y.-P., Goeddel, D.V., Giles, K.M., Pun, (1996a). Systemic injection of a tripeptide inhibits the intracellular
K.-T., Grinham, C.J., Brown, R., and Farrow, S.N. (1996). A death- activation of CPP32-like proteases in vivo and fully protects mice
domain-containing receptor that mediates apoptosis. Nature 384, against Fas-mediated fulminant liver destruction and death. J. Exp.
372–375. Med., in press.
Knudson, C.M., Tung, K.S., Tourtellotte, W.G., Brown, G.A., and Rodriguez, I., Matsuura, K., Khatib, K., Reed, J.C., Nagata, S., and
Review: Apoptosis by Death Factor
365
Vassalli, P. (1996b). A bcl-2 transgene expressed in hepatocytes Bcl-2-deficient mice demonstrate fulminant lymphoid apoptosis,
polycystic kidneys, and hypopigmented hair. Cell protects mice from fulminant liver destruction but not from rapid 75, 229–240.
death induced by anti-Fas antibody injection. J. Exp. Med. 183, Wang, C.-Y., Mayo, M.W., and Balwin Jr, A.S. (1996). TNF- and
1031–1036. cancer therapy–induced apoptosis: potentiation by inhibition of NFRothe,
M., Wong, S.C., Henzel, W.J., and Goeddel, D.V. (1994). A kB. Science 274, 784–787.
novel family of putative signal transducers associated with the cyto- Watanabe-Fukunaga, R., Brannan, C.I., Copeland, N.G., Jenkins,
plasmic domain of the 75 kDa tumor necrosis factor receptor. Cell N.A., and Nagata, S. (1992). Lymphoproliferation disorder in mice
78, 681–692. explained by defects in Fas antigen that mediates apoptosis. Nature
Rothstein, T.L., Wang, J.K.M., Panka, D.J., Foote, L.C., Wang, Z., 356, 314–317.
Stanger, B., Cui, H., Ju, S.-T., and Marshak-Rothstein, A. (1995). Wu, J., Zhou, T., Zhang, J., He, J., Gause, W.C., and Mountz, J.D.
Protection against Fas-dependent Th1-mediated apoptosis by anti- (1994). Correction of accelerated autoimmune disease by early regen
receptor engagement in B cells. Nature 374, 163–165. placement of the mutated lpr gene with the normal Fas apoptosis
Schulze-Osthoff, K., Walczak, H., Dro¨ ge, W., and Krammer, P.H. gene in the T cells of transgenic MRL-lpr/lpr mice. Proc. Natl. Acad.
(1994). Cell nucleus and DNA fragmentation are not required for Sci. USA 91, 2344–2348.
apoptosis. J. Cell Biol. 127, 15–20. Yang, E., and Korsmeyer, S.J. (1996). Molecular thanatopsis: a disShaham,
S., and Horvitz, H.R. (1996). Developing Caenorhabditis course on the Bcl2 family and cell death. Blood 88, 386–401.
elegans neurons may contain both cell-death protective and killer Yonehara, S., Ishii, A., and Yonehara, M. (1989). A cell-killing monoactivities.
Genes Dev. 10, 578–591. clonal antibody (anti-Fas) to a cell surface antigen co-downregulated
Singer, G.G., and Abbas, A.K. (1994). The Fas antigen is involved in with the receptor of tumor necrosis factor. J. Exp. Med. 169, 1747–
peripheral but not thymic deletion of T lymphocytes in T cell receptor 1756.
transgenic mice. Immunity 1, 365–371. Yuan, J.,Shaham, S., Ledoux, S., Ellis, H.M., and Horvitz, H.R. (1993).
Spiegel, S., Foster, D., and Kolesnick, R. (1996). Signal transduction The C. elegans cell death gene ced-3 encodes a protein similar to
through lipid second messengers. Curr. Opin. Cell Biol. 8, 159–167. mammalian interleukin-1b-converting enzyme. Cell 75, 641–652.
Strand, S., Hofmann, W.J., Hug, H., Mu¨ ller, M., Otto, G., Strand, D., Zamzami, N., Susin, S.A., Marchetti, P., Hirsch, T., Go´ mez-MonterMariani,
S.M., Stremmel, W., Krammer, P.H., and Galle, P.R. (1996). rey, I., Castedo, M., and Kroemer, G. (1996). Mitochondrial control
Lymphocyte apoptosis induced by CD95 (APO-1/Fas) ligand ex- of nuclear apoptosis. J. Exp. Med. 183, 1533–1544.
pressing tumor cells—A mechanism of immune evasion. Nature Zha, J., Harada, H., Yang, E., Jockel, J., and Korsmeyer, S.J. (1996).
Med. 2, 1361–1366. Serine phosphorylation of death agonist BAD in response to survival
Suda,T., Takahashi, T., Golstein, P., andNagata, S. (1993). Molecular factor results in binding to 14–3–3 not Bcl-x. Cell 87, 619–628.
cloning and expression of the Fas ligand: a novel member of the
tumor necrosis factor family. Cell 75, 1169–1178. Note Added in Proof
Takahashi, A., Alnemli, E.S., Lazebnik, Y.A., Fernandes-Alnemri, T.,
Litwack, G., Moir, R.D., Goldman, R.D., Poirier, G.G., Kaufmann, The paper cited as Rodriguez et al. (1996a) has been published:
S.H., and Earnshaw, W.C. (1996). Cleavage of lamin A by Mch2a Rodriguez, I., Matsuura, K., Ody, C., Nagata, S., and Vassalli, P.
but not CPP32: Multiple interleukin 1b–converting enzyme-related (1996a). Systemic injection of a tripeptide inhibits the intracellular
proteases with distinct substrate recognition properties are active activation of CPP32-like proteases in vivo and fully protects mice
in apoptosis. Proc. Natl. Acad. Sci. USA 93, 8395–8400. against Fas-mediated fulminant liver destruction and death. J. Exp.
Med. 184, 2067–2072.
Takahashi, T., Tanaka, M. Brannan, C.I., Jenkins, N.A., Copeland,
N.G., Suda, T. and Nagata, S. (1994). Generalized lymphoproliferative
disease in mice, caused by a point mutation in the Fas ligand.
Cell 76, 969–976.
Tanaka, M., Suda, T., Haze, K., Nakamura, N., Sato, K., Kimura, F.,
Motoyoshi, K., Mizuki, M., Tagawa, S., Ohga, S., et al. (1996). Fas
ligand in human serum. Nature Med. 2, 317–322.
Tanaka, M., Suda, T., Yatomi, T., Nakamura, N., and Nagata, S.
(1997). Lethal effect of recombinant human Fas ligand in mi