(Received for publication, May 9, 1997)
From the Center for Apoptosis Research and the Department of Microbiology and Immunology, Kimmel Cancer Institute, Thomas Jefferson University, Philadelphia, Pennsylvania 19107, ¶ Center for Apoptosis Research and the Department of Biochemistry and Molecular Pharmacology, Thomas Jefferson University, Philadelphia, Pennsylvania 19107, and § IDUN Pharmaceuticals, San Diego, California 92121
We identified and cloned a novel human protein that contains FADD/Mort1 death effector domain homology regions, designated FLAME-1. FLAME-1, although most similar in structure to Mch4 and Mch5, does not possess caspase activity but can interact specifically with FADD, Mch4, and Mch5. Interestingly, FLAME-1 is recruited to the Fas receptor complex and can abrogate Fas/TNFR-induced apoptosis upon expression in FasL/tumor necrosis factor-sensitive MCF-7 cells, possibly by acting as a dominant-negative inhibitor. These findings identify a novel endogenous control point that regulates Fas/TNFR1-mediated apoptosis.
Several members of the caspase family of proteases (1) have been implicated as key regulators of programmed cell death or apoptosis (2, 3). The proapoptotic caspases can be divided into two groups: those with a large prodomain such as ICH-1 (caspase-2), Mch4 (caspase-10), Mch5/MACH/FLICE (caspase-8) and Mch6/ICE-Lap-6 (caspase-9) and those with a small prodomain such as CPP32/YAMA/Apopain (caspase-3), Mch2 (caspase-6), and Mch3/ICE-Lap-3 (caspase-7). Caspases with large prodomains are probably the most upstream caspases (4, 5). They are recruited by several death-signaling receptors that belong to the TNFR family (6) through interactions of their prodomain with the receptor-interacting adaptor molecules FADD/Mort1 (7, 8) or CRADD/RAIDD (5, 9). For example, the prodomains of Mch4 and Mch5 contain two tandem regions that show significant homology with the N-terminal death effector domain (DED)1 of FADD (10-12). Engagement of Fas/TNFR1 results in recruitment of FADD to the receptor complex (13), which presumably triggers activation of the caspase apoptotic pathway through interaction of its DED with the corresponding motifs in the prodomain of Mch5 and probably Mch4. CRADD presumably functions like FADD by recruiting ICH-1 to the Fas/TNFR1 complex, through interaction of its N-terminal domain with the corresponding motif in the prodomain of ICH-1 (5, 9). Thus, the prodomains of caspases function to physically link the death receptors to the downstream caspase activation pathway.
In our efforts to characterize novel apoptotic/anti-apoptotic molecules that contain FADD-DED homology regions, we have identified a human molecule designated FLAME-1 (FADD-like antiapoptotic molecule) that is structurally related to Mch4 and Mch5. We show that FLAME-1 is recruited to the Fas signaling pathway through interaction with FADD. Interestingly, FLAME-1 can block Fas/TNFR1-induced apoptosis but not UV-induced apoptosis.
The full-length FLAME-1 cDNA was
cloned from human Jurkat Uni-ZAP XR cDNA library (14) by screening
the library with a partial FLAME-1 cDNA probe. The probe was
amplified by PCR using FLAME-1-specific primers derived from the 3
(GenBank accession no. aa002262) and 5
(GenBank accession no.
aa001257) sequences of human expressed sequence tag (EST) clone
427786.
The human genes for Mch5 and FLAME-1 were mapped on previously described rodent-human hybrid panels (15) and on the Genebridge 4 and Stanford G3 radiation hybrid panels (Research Genetics) using specific oligonucleotide primers.
In Vitro Binding AssaysThis was performed as described recently (5).
Mammalian Expression VectorsEpitope tagging was done by cloning cDNAs in-frame into the multiple cloning sites of a modified T7-pcDNA3 and/or the Flag plasmid pFLAG-CMV-2 (IBI Kodak). All deletion and point mutants were generated by PCR and verified by sequencing. Flag-tagged Fas was constructed in pcDNA3 as described (8).
Transfection, Immunoprecipitation, and Western Analysis293 or 293T human embryonic kidney cells were transiently transfected with the expression plasmids using the LipofectAMINE (Life Technologies, Inc.) method. Cell lysis, immunoprecipitation, and Western blotting were done as described (9).
Yeast Two-hybrid AnalysisMch4-FDH (residues 18-189),
Mch5 (FDH A, residues 3-80; FDH B, residues 102-177; FDH, residues
3-177), FLAME-1-FDH (residues 1-160), and murine FADD-DED (residues
1-78) were subcloned into yeast two-hybrid vectors. Yeast two-hybrid
analysis was then performed as described recently (16).
MCF7-FAS cells were transiently cotransfected with reporter and test plasmids at a ratio of 1:10 and assayed for apoptosis as described (5, 16). The percentage of viable cells (mean ± S.D.) under each condition was determined by measuring the number of viable blue cells compared with total blue cells.
An EST (clone 427786)
with statistically significant similarity to Mch5 (p < 0.001) was identified. Based on its sequence a probe was generated by
PCR and used to isolate and clone the full-length cDNA (~2 kb)
from a human Jurkat cDNA library. This cDNA encodes a novel
protein (designated FLAME-1) of 445 amino acids with
predicted relative molecular mass of 51 kDa (Fig.
1A).
Sequence analysis, tissue distribution, and
chromosomal localization. A, colinear alignment of the
predicted amino acid sequence of human FLAME-1 with pro-Mch4 and
pro-Mch5 and a schematic diagram of its structure. Based on the
crystal structure of ICE and CPP32, the residues marked with a
c are involved in catalysis and those marked with a
b are involved in binding the carboxylate side chain of the
substrate P1 aspartate. The active site pentapeptide QACQG in Mch4 and
Mch5 is boxed. The residues that are unique to FLAME-1
are underlined. The vertical arrow indicates the
splice junction, after which FLAME-1
differs from FLAME-1.
Boldface denotes identical residues. B, Northern
blot analysis of FLAME-1 mRNAs. Numbers on the right indicate
kilobases. PBL, peripheral blood leukocyte. C,
FLAME-1, Mch4, and Mch5 genes are localized to chromosome
2q33-34.
Structural Organization of FLAME-1
FLAME-1 is most similar to
the Mch4 and Mch5 caspases (Fig. 1A). It has three distinct
homology regions; there are two N-terminal tandem stretches of
approximately 67-79 residues that are significantly homologous to the
N-terminal DED (residues 1-79) of FADD, here referred to as FADD-DED
homology A (FDH-A, residues 5-71) and B (FDH-B, residues 90-168)
regions. FDH-A and FDH-B share 38 and 28% identity with the DED of
FADD, respectively. The FDH regions share 28-33% identity with the
corresponding regions in Mch4 and Mch5. They are followed by a
stretch of 249 residues (residues 197-445) with significant homology
to the caspase domain of Mch4 or Mch5 (27-31% identity), here
referred to as the caspase domain homology (CDH) region. The CDH
contains a QNYVV motif instead of the conserved active site motif
QACXG (where X is R,Q,G) present in caspases.
Also, only one (Gly-281) out of the three residues involved in
catalysis and two (Gln-323 and Ser-386) out of the four residues
involved in binding the carboxylate side chain of the substrate P1
aspartate are conserved (Fig. 1A). Interestingly, this
region contains a potential caspase cleavage site (LEVD
G) C-terminal
to the QNYVV motif that can be cleaved by caspases (see below) to
generate two polypeptides (p39 and p12) corresponding to the large and
small subunits of caspases. These observations suggest that FLAME-1
could be a protease with a different substrate specificity compared
with caspases or an enzymatically inactive protein. A naturally
existing alternatively spliced isoform of FLAME-1 (FLAME-1
) lacking
the entire CDH region was also identified by reverse transcription-PCR.
This isoform shares residues 1-231 with FLAME-1 but has a 39-amino
acid-long unique C terminus (Fig. 1A).
Northern blot analysis (Fig. 1B) revealed that FLAME-1 mRNA (~2 kb) is expressed mainly in testes and skeletal muscle. This message is less abundant in the other human tissues examined. However, a ~1.2-kb abundant message is expressed in the placenta, which could be an alternatively spliced isoform of FLAME-1 mRNA.
Chromosomal Localization of FLAME-1Chromosomal mapping linked the FLAME-1 and Mch5 genes to the D2S116 and D2S348 markers on chromosome 2q33-34 using radiation hybrid panels, in close proximity to where we had previously localized Mch4 (10) (Fig. 1C). This finding and the high degree of homology among their genes or gene products suggest that they might be descendents of a common ancestral gene through gene duplication.
FLAME-1 Is a Target of Active CaspasesUnlike Mch4 or Mch5
(4), expression of FLAME-1 in bacteria or in the baculovirus expression
system did not result in its cleavage (autoactivation) or generation of
a caspase-like activity as determined with the tetrapeptide substrates
YVAD-AMC (where AMC is 7-amino-4-methylcoumarin) or DEVD-AMC (data not
shown), suggesting that FLAME-1 might be enzymatically inactive or
possess an unknown enzymatic activity. Interestingly, in
vitro translated FLAME-1 can be cleaved by several purified
caspases including CPP32, Mch2, Mch3, Mch4, Mch5, and ICH-1 to generate
two fragments (p39 and p12) topologically equivalent to the large and
small subunits of caspases (Fig.
2A). This cleavage occurs at
Asp-341 in the LEVDG site, since a D to A mutation in this site
prevents these caspases from cleaving FLAME-1. Transfection studies
showed that FLAME-1 may also be a caspase substrate in vivo.
Expression of a T7 epitope-tagged FLAME-1 (T7-FLAME-1) in 293 cells
produced both full-length and cleaved (p39) FLAME-1 (see Fig. 3,
B and E). This
cleavage was not observed with the D341A mutant FLAME-1 (T7-FLAME-1-D341A, Fig. 3, C and E). Furthermore,
stimulation of FLAME-1-transfected MCF7-FAS cells with anti-Fas
antibody increased the amount of cleavage products, whereas addition of
caspase inhibitors significantly reduced it (not shown). Thus, FLAME-1
appears to be a caspase target in apoptotic cells.
Interactions of FLAME-1
To investigate the participation of
FLAME-1 in Fas/TNFR1 apoptotic signaling pathways, in vitro
and in vivo binding studies and yeast two-hybrid analysis
were performed. Radiolabeled FLAME-1, Mch4, Mch5, FADD, or mutants
of these proteins were precipitated with various glutathione
S-transferase (GST) fusion proteins immobilized on
glutathione-Sepharose beads (Fig. 2, B-D). Mch4, Mch5
,
and FLAME-1 associated specifically with GST-FADD, although the
interaction of FLAME-1 with FADD was weaker than that observed with
Mch4 or Mch5
(Fig. 2B). FADD, FADD-DED, Mch4, Mch5
,
and Mch5
-FDH, but not FADD-DD, also associated specifically with
FLAME-1
(GST-FLAME-1
) (Fig. 2C). These observations
suggest that the interactions are mediated by the homologous FDH
regions of these proteins. Interestingly, Mch5
but not Mch4
associated with a truncated FLAME-1 lacking its FDH regions
(GST-FLAME-1-CDH), suggesting that the two proteins can also interact
through their homologous CDH regions (Fig. 2D).
To demonstrate these interactions in vivo, we transiently
transfected 293 cells with plasmids encoding T7 epitope-tagged FADD, FLAME-1, or mutants and various Flag epitope-tagged proteins. Because
wild type Mch4, Mch5, and their CDH regions are potent inducers of
apoptosis in 293 cells (not shown), active site Cys to Ala Flag-tagged
mutants were used in these experiments to investigate their
interactions with FLAME-1. Consistent with the in vitro results, T7-FADD coprecipitated with full-length FLAME-1, Mch4, and
Mch5
or their isolated FDH regions (Fig. 3A). T7-FLAME-1 and its p39 fragment coprecipitated with full-length FADD, Mch4, Mch5
, or their isolated FDH regions (Fig. 3B).
Full-length T7-FLAME-1 (but a negligible amount of the p39 fragment)
associated with Mch5-CDH (Fig. 3B), suggesting that the
entire CDH region of FLAME-1 is required for optimal interaction
between these proteins. Similar results were obtained with
T7-FLAME-1-D341A and T7-FLAME-1-CDH (Fig. 3, C and
D). No interactions were observed between T7-FLAME-1
and
Flag-Mch4-CDH or Flag-Mch5
-CDH (data not shown), suggesting that
these proteins can only interact through their respective FDH or CDH
regions. The yeast two-hybrid analysis confirmed the interactions of
FLAME-1-FDH with FADD, Mch4, and Mch5 FDH regions (Table
I). This analysis also revealed that
FLAME-1-FDH can also strongly interact with itself (Table I).
|
FADD can recruit Mch5 (MACH/FLICE) (11, 12) and possibly Mch4 (10) to the Fas/TNFR1 signaling complex. To determine whether FLAME-1 can also be recruited through FADD, coprecipitation experiments were performed in 293T cells (Fig. 3E). FLAME-1 or FLAME-1-D341A was able to form a complex with Fas (lanes 4 and 8), possibly through interaction with endogenous FADD. Cotransfection of exogenous T7-FADD enhanced the FLAME-1-Fas interaction (lanes 5 and 7). The p39 fragment, which is generated by cleavage at Asp-341, also formed a complex with Fas (lane 5). Interestingly, FLAME-1 was also able to prevent recruitment of Mch4 and Mch5 to Fas (not shown). These observations demonstrate that FLAME-1 can be recruited to Fas and may interfere with the assembly of a functional death signaling complex.
Inhibition of Fas/TNF-induced Apoptosis by FLAME-1To study
the functional role of FLAME-1 in Fas/TNFR1- or UV-induced apoptosis,
it was transfected into MCF-7-FAS cells. FLAME-1 did not induce
apoptosis in these cells (Fig. 4).
However, FLAME-1 and FLAME-1 significantly blocked Fas- and
TNFR1-induced apoptosis but not UV-induced apoptosis (Fig. 4). This
indicates that overexpression of the FDH regions of FLAME-1 is
sufficient to block Fas/TNFR1-induced apoptosis. This protective effect
approached 60-65% of that observed by Bcl-xL overexpression. The
isolated CDH region of FLAME-1 did not have any protective effect (not
shown).
Taken together, the data presented here establish FLAME-1 as the first example of an endogenous FDH-containing protein that can act as a negative regulator of apoptosis. Recently, we demonstrated that viral FDH-containing proteins E8 and MC159 can abrogate Fas/TNFR1-mediated apoptosis (16). Both FLAME-1 and the viral proteins appear to target the Fas/FADD/caspase signaling complex by a potential dominant negative mechanism. Binding of FLAME-1, its FDH regions, or the viral proteins to the caspases Mch4 or Mch5 or the adaptor molecule FADD blocks Fas/TNFR1-induced apoptosis possibly by interfering with the assembly of a functional death receptor signaling complex. Consequently, it appears that molecules which contain FDH regions could be either proapoptotic like FADD, Mch4, or Mch5 or anti-apoptotic such as FLAME-1 and the viral proteins E8 and MC159. Since the proapoptotic or anti-apoptotic proteins might have different expression levels, their ratios could determine how a given cell or cell type responds to FasL or TNF. For example, the high expression of FLAME-1 in Fas+, FasL+ immune privileged tissues (i.e. testis and placenta, see Fig. 1B) could be the reason for their resistance to FasL killing.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) AFOO9616-AFOO9620.