F1Aα, a Death Receptor-binding Protein Homologous to theCaenorhabditis elegans Sex-determining Protein, FEM-1, Is a Caspase Substrate That Mediates Apoptosis*

Apoptosis is an evolutionarily conserved process that is critical for tissue homeostasis and development including sex determination in essentially all multicellular organisms. Here, we report the cloning of an ankyrin repeat-containing protein, termed F1Aα, in a yeast two-hybrid screen using the cytoplasmic domain of Fas (CD95/APO-1) as bait. Amino acid sequence analysis indicates that F1Aα has extensive homology to the sex-determining protein FEM-1 of the Caenorhabditis elegans, which is required for the development of all aspects of the male phenotype. F1Aα associates with the cytoplasmic domains of Fas and tumor necrosis factor receptor 1, two prototype members of the “death receptor” family. The F1Aα protein also oligomerizes. Overexpression of F1Aα induces apoptosis in mammalian cells, and co-expression of Bcl-XL or the dominant negative mutants of either FADD or caspase-9 blocks this effect. Deletion analysis revealed the center region of F1Aα, including a cluster of five ankyrin repeats to be necessary and sufficient for maximum apoptotic activity, and the N-terminal region appears to regulate negatively this activity. Furthermore, F1Aα is cleaved by a caspase-3-like protease at Asp342, and the cleavage-resistant mutant is unable to induce apoptosis upon overexpression. F1Aα is therefore a member of a growing family of death receptor-associated proteins that mediates apoptosis.

Caspases are critical mediators of apoptosis (5)(6)(7)(8). In addition to autoactivation and activation of other caspases, caspases are thought to participate in apoptosis by disabling important cellular processes and breaking down structural components of the cell. Caspases also activate signaling molecules that upon cleavage commit the cells to apoptosis. Molecules that transmit death signals upon cleavage by caspases have been identified in various apoptotic pathways (9 -13).
Some members of the tumor necrosis factor (TNF) 1 receptor superfamily, known as the death receptors (14 -17), efficiently transmit death signals via a cytoplasmic motif called the "death domain" (18,19). Among members of the death receptor family, the receptor-proximal events have been best characterized for Fas and TNFR1. Stimulation of these receptors results in aggregation of their intracellular death domains (20,21), leading to the recruitment of key signaling proteins (14). The Fas receptor signals to caspase-8 through the recruitment of the adaptor protein, FADD/MORT1 (22)(23)(24)(25), whereas TNFR1 signals to caspase-8 and caspase-2 through the TRADD-FADD (26) and TRADD-RIP-RAIDD/CRADD (27,28) pathways, respectively. Although the precise mechanism is still not clear, it is known that formation of a death receptor-FADD-caspase 8 complex is required for the activation of caspase-8, which is an early step in one of the cascades of apoptotic events induced by Fas and TNFR1 (10,11,13,29,30).
In our effort intended to identify additional components of the death receptor signaling pathways, we identified an ankyrin-repeat containing protein, termed F1A␣, in a yeast two-hybrid screen using the cytoplasmic domain of the mouse Fas receptor (mFas) as bait. The amino acid sequence of F1A␣ is highly homologous to the C. elegans protein, FEM-1, which is essential for achieving all aspects of the male phenotype in the nematode (31). F1A␣ binds to mFas, TNFR1, and itself. Overexpression of F1A␣ induces apoptosis in MCF7 cells that can be blocked by expression of Bcl-XL or the dominant negative mutants of either FADD or caspase-9. F1A␣ is therefore a member of a growing family of death receptor-associated proteins (25,(32)(33)(34)(35) that mediate apoptosis.

EXPERIMENTAL PROCEDURES
Reagents and Cell Lines-Mono-and polyclonal antibodies against the Myc epitope (9E10, A14) and polyclonal antibody against the HA epitope (Y11) were obtained from Santa Cruz Biotechnology. Monoclonal antibody against the HA epitope (12CA5) was purchased from Roche Molecular Biochemicals. Antibody against PARP (C2-10) was from Dr. G. Poirier, CHUL Research Center, Canada. HeLa, 293, and NIH3T3 were originally from American Type Culture Collection (ATCC). The TNF-sensitive MCF7 (breast carcinoma cells) was provided by Dr. V. Dixit, University of Michigan. Cell lines were grown according to the directions provided by suppliers. All media were supplemented with 10% heat-inactivated fetal bovine serum (Life Technologies, Inc.) and antibiotics (100 mg of streptomycin/ml and 100 IU of penicillin/ml, Life Technologies, Inc.). The peptide protease inhibitors YVAD-fmk, Z-VAD-fmk and DEVD-fmk were from Enzyme System Products, CA. USA. Recombinant caspase-3, -6, -7, and 8 were purchased from PharMingen.
Plasmids Construction-Plasmids containing the cDNA for TNFR1 and NGFR were from M. Chao, Cornell University Medical College and mFas from K. B. Elkon, Cornell University. DNA fragments for plasmid constructs containing mFas-IC-(166 -306), mFas-FD5-(166 -292), TNFR1-IC-(205-426), TNFR1-IC⌬15-(205-411), and NGFR-IC-(271-427) were obtained by PCR amplification using the Expand TM high fidelity polymerase chain reaction (PCR) System (Roche Molecular Biochemicals) with primers incorporated with appropriate restriction sites and epitope tags as needed into the pXJ40 mammalian expression vector driven by the CMV promoter (36). F1A␣ expression vector was constructed by PCR amplification of cDNA fragment from the fulllength positive clone obtained in the cDNA screen. The PCR primers were incorporated with appropriate restriction sites for insertion of the amplified fragment such that it was in-frame with the sequences encoding the Myc-or HA epitope in the pXJ40 mammalian expression vector driven by the CMV promoter (36). The constructs were sequenced to ensure that no PCR error was introduced. Unless otherwise stated all epitope tags are at the N termini. The N-terminal deletion mutants of F1A␣ were generated by PCR and re-ligation of appropriate 5Ј end fragment to the original F1A␣ construct to avoid mutation introduced by PCR error. DNA fragments generated by PCR and the junctions of insertion were confirmed by sequencing. The C-terminal deletion mutants of F1A␣ were made by introducing a stop codon mutation at the indicated position, and point mutations were performed using the Transformer TM Site-directed Mutagenesis Kit (CLONTECH). cDNA encoding Bcl-XL, and caspase-9 were obtained from Craig B. Thompson, The University of Chicago, and Peng Li, Institute of Molecular and Cell Biology, Singapore, respectively. FADD-DN (FADD-(80 -208)) was obtained by PCR and caspase-9-DN (caspase-9-C287A) was generated by site-directed mutagenesis.
Yeast Two-hybrid Cloning-The methodology and reagents used for the yeast two-hybrid cloning were essentially the same as described in Bai and Elledge (37). HB12 was cloned from a GAL4 AD-tagged human B cell library kindly provided by Dr. S. Elledge, Baylor College of Medicine. Briefly, the cytoplasmic domain of the mouse Fas antigen and mFas-FD5-(166 -292) were obtained by PCR and cloned in-frame, as confirmed by sequencing, into the GAL4 DNA-binding domain vector pAS1-CYH2. Screening was performed according to the Matchmaker Two-hybrid System Protocol (CLONTECH) in the presence of 30 mM 3-aminotrizole. The binding properties of HB12, as well as of the other examined proteins, were assessed in the yeast Y190 reporter strain. Filter lift assays for ␤-galactosidase activity were performed to detect interaction between fusion proteins. The following heterologous proteins were expressed in GAL4 DNA-binding domain vector to test for possible interaction with the putative positive clones isolated from the primary screen: p53, retinoic acid receptor (hRAR␣), retinoid X receptor (hRXR␣), SNF, lamin, and CDK2.
cDNA Library Screening-Two cDNA fragments of approximately 400 base pairs from the 5Ј and 3Ј ends of HB12 were labeled with [ 32 P]dATP and used to screen human pancreas and spleen gt11 cDNA libraries (CLONTECH) using standard techniques as described (38). 10 6 clones of each libraries were screened. Positive clones were isolated and the cDNA inserts from these clones were subcloned into the pB-SK(II) vector (Stratagene) and characterized.
Northern Blot Analysis-The human and mouse multiple tissue Northern blots (CLONTECH) were hybridized with a 32 P-labeled cDNA insert obtained from an XhoI digestion of Gal4ad-HB12 clone, using the ExpressHyb TM Hybridization Solution (CLONTECH) according to the instructions of the manufacturer.
In Vitro Binding Analysis-Sequences encoding mFas-IC, mFas-FD5, TNFR1-IC, TNFR1-IC⌬15, FADD, and retinoid X receptor were excised from pAS vector as NdeI-SalI or NcoI-SalI fragments and cloned in-frame into GST fusion protein vector pGEX-TK4E (38). The plasmids were transformed into the Escherichia coli strain BL 21. GST and GST fusion protein were prepared by standard methods (38), and the recombinant proteins were immobilized onto glutathione-agarose beads. Labeled F1A␣ was prepared by in vitro transcription/translation of pXJ-HA-F1A␣ using TNT T7-coupled reticulocyte lysate system from Promega. The integrity of the 35 S-labeled proteins was verified by SDS-PAGE. For in vitro protein interaction, equal amounts of total 35 S-labeled lysate (500,000 cpm of trichloroacetic acid-precipitable counts) were diluted into 1 ml of GST binding buffer (50 mM Hepes (pH 7.6), 5 mM EDTA, 250 mM NaCl, 0.1% Nonidet P-40) and incubated for 1 h with the various GST fusion proteins immobilized on the beads (approximately 2 g). Samples were subsequently washed 6 times with binding buffer and boiled for 3 min in loading buffer before fractionation on 10% SDS-polyacrylamide gel electrophoresis (SDS-PAGE). Bound proteins were visualized by autoradiography.
Immunoprecipitation-For co-immunoprecipitation experiments, 293 cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum to 80% confluency. Transfection was carried out with LipofectAMINE (Life Technologies, Inc.) according to the manufacturer's instructions. Cells were incubated with the Li-pofectAMINE/DNA mixture for 12 h followed by change of fresh media. The cells were harvested 8 h after change of media and lysed in 1 ml of lysis buffer (50 mM HEPES (pH 7.6), 350 mM NaCl, 1% Nonidet P-40, and 5 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 50 g/ml aprotinin, and 10 g/ml leupeptin). An aliquot (10 l) of the cell lysates was fractionated on SDS-PAGE for visualization of the expression of proteins. The remaining cell lysates were incubated with 1 g of polyclonal anti-HA antibody for 1 h on ice and then mixed with 20 l of a 1:1 slurry of protein A-agarose and incubated for another 1 h at 4°C. The agarose beads were washed once in 1 ml of lysis buffer, 2 times in 1 ml of lysis buffer containing 500 mM NaCl, and 2 times in 1 ml of lysis buffer before fractionation on SDS-PAGE. Western blotting analyses were performed subsequently with procedure as described previously (39) using monoclonal anti-Myc antibody.
Apoptosis Assays-MCF7 cells were maintained in a 35-mm dish in RPMI media supplemented with fetal bovine serum and transfected using LipofectAMINE. Transfections were carried out in 1 ml of serumfree Dulbecco's modified Eagle's medium and incubated for 6 h after which 1 ml of serum-containing RPMI was added. ␤-Galactosidase assays were performed on the cells 18 h later which was 24 h posttransfection or 48 h post-transfection as indicated. The cells were washed once with PBS, fixed with 2% formaldehyde and 0.2% glutaraldehyde in 1ϫ PBS for 5 min at 4°C, and stained with a histochemical reaction mixture (5 mM potassium ferricyanide, 5 mM potassium ferrocyanide, 2 mM MgCl 2 , and 1 mg/ml X-gal) at 37°C. After 16 h, cells were visualized by phase-contrast microscopy. For detection of PARP cleavage, MCF7 cells cultured on 100-mm dishes were transiently transfected, and nuclear extracts were prepared as described (39). The extracts were fractionated on SDS-PAGE followed by Western blotting analyses using PARP-specific antibody (C2-10).
Nuclear Staining Assay-The assay was performed with procedure as described previously (40). Briefly, cells were seeded onto glass coverslips and transfected with plasmids expressing F1A␣, its deletion mutants, or vector control. 24 h after transfection, the cells in monolayers were washed twice with ice-cold PBS (pH 7.4) and fixed for 5 min at 4°C with absolute methanol (Ϫ20°C). The washing step with PBS was then repeated once. To stain the nuclei, the cells were incubated for 10 min with 10 g/ml Hoechst 33342 (Molecular Probes Inc., Eugene, OR) and then washed with PBS. The coverslips with the stained cells were mounted in 80% glycerol in PBS containing 1 mg/ml p-phenylenediamine and examined with a Zeiss Axioplan microscope.

RESULTS
Cloning of HB12-A yeast two-hybrid screen was performed to isolate clones that may interact with the cytoplasmic domain of the mouse Fas receptor (mFas). The cDNA fragment encoding amino acids 166 -292 of mFas (mFasFD5) was cloned inframe with the DNA binding domain of Gal4 (GBD-mFasFD5) and used to screen a human B cell library. A truncated cytoplasmic domain of Fas (mFasFD5) was used because deletion of a negative regulatory region at the C terminus of Fas markedly enhances its ability to mediate apoptosis (18) and to bind FADD (25). Approximately 3 million independent transformants were screened, and 17 putative positive clones were identified, of which two failed to show interaction with any of the heterologous proteins (see "Experimental Procedures"). Sequence analysis revealed that one of the clones encoded the full-length FADD/MORT1 (18,24), whereas the other clone, HB12, contains a partial complementary DNA with an open reading frame encoding a polypeptide of 469 amino acids.
Although the HB12 clone encoded an N-terminal truncated protein, the 469 amino acid polypeptide may contain the effector domain of the full-length protein. Since HB12 interacted with the death receptors, its ability to modulate apoptosis was evaluated by overexpressing HB12 in MCF7 cells by transient transfection. MCF7 cells were co-transfected with pCMV-␤galactosidase as a marker for transfected cells and expression vector containing clone HB12 or control vector. In the plate transfected with the HB12 clone, a significant percentage of total blue (␤-galactosidase-positive) cells displayed the round cell morphology typical of cells that are undergoing apoptosis (Fig. 1A). Chromatin condensation, another hallmark of apop-tosis, was evident among the nuclei of the round cells when they were stained with the Hoechst 33342 fluorescent dye (Fig.  1B). Proteolytic cleavage of poly(ADP-ribose) polymerase (PARP), which serves as a marker for the activation of caspases in cells undergoing apoptosis (41), was also observed in cells transfected with HB12 expression vector (Fig. 1C). HB12-induced cell death was inhibited by the caspase inhibitors Z-VAD-fmk and DEVD-fmk (Fig. 1D). The cleavage of PARP to the signature 85-kDa apoptotic fragment in the MCF7 cells transfected with HB12 was completely blocked by treatment of the cells with 20 M Z-VAD-fmk or DEVD-fmk (Fig. 1C).
Isolation of F1A␣-By using HB12 cDNA fragment as a probe, we screened a human hypothalamus cDNA library (CLONTECH) and obtained several cross-hybridizing cDNA clones. One of these, clone HH6.15, contained a 3.1-kb cDNA insert with a 1881-nucleotide open reading frame beginning with a translational initiation consensus sequence (42) and predicted a protein of 627 amino acids with a molecular mass of 70 kDa (Fig. 2). A corresponding murine cDNA clone was subsequently obtained by screening a mouse testis library. The amino acid sequence deduced from the murine cDNA clone is virtually identical to that of the human clone ( Fig. 2A). Data base searches revealed that the predicted protein shares extensive similarity throughout the entire sequence, including the presence of six tandemly arranged ankyrin-repeats at the N terminus (Fig. 2B), with the sex-determining protein in C. elegans known as FEM-1 ( Fig. 2A). Because of its ability to modulate apoptosis in mammalian cells we named the protein F1A for FEM-1-like protein in the apoptotic pathway.
To determine the tissue distribution of F1A mRNA, Northern analysis was performed with the HB12 cDNA as a probe. Two transcripts, 5.5 and 7.5 kb, were found to be ubiquitously distributed, with varying abundance; two additional smaller transcripts, 1.35 and 2.5 kb, were found only in the testis (Fig.  2C). In the mouse tissues, only the ubiquitously distributed 7.5-kb and the testis-specific 2.4-kb transcripts were detected (data not shown). The different transcripts may be generated by alternate splicing or derived from other related genes. Sequence analysis of several independent partial cDNA clones failed to detect the existence of splice variants or related family members. However, a search in the EST data base from NCBI has revealed several human and murine cDNA clones that share high level of similarity (Ͼ30%) in amino acid sequence with F1A, suggesting F1A might be a member of a gene family. During the preparation of this manuscript, the deduced amino acid sequence of two members, Fem1a and Fem1b, of a mouse gene family was reported (43). The amino acid sequence of Fem1b is identical to that of mF1A confirming that F1A is indeed a member of a gene family. We therefore refer to our protein as F1A␣.
F1A␣ Specifically Interacts with the Death Receptors and Self-associates-In the two-hybrid assay, clone HB12 (F1A␣-(159 -627)) interacted with mFas-FD5-(166 -292) and the intracellular domain of TNFR1 (TNFR1-IC-(205-426)). Radiolabeled in vitro translated full-length F1A␣ was tested for in vitro binding with various glutathione S-transferase (GST) fusion proteins (Fig. 3A). F1A␣ specifically associated with GST-mFas-FD5 and GST-TNFR1-IC in the assay but not with either GST or GST-FADD. In contrast to the two-hybrid results (data not shown), F1A␣ interacted equally well with both GST-mFas-IC-(166 -306) and GST-mFas-FD5. Parallel experiments using HB12 clone, F1A␣-(159 -627), yielded identical results (data not shown). In vitro self-association of F1A␣ was not evaluated because we were unable to obtain a reasonable yield of GST-F1A␣; however, 35 S-F1A␣ was found to associate with GST-F1A␣-(482-627) (data not shown). To prevent this, Z-VAD-fmk at 20 M was added to the culture media. HA-F1A␣ was immunoprecipitated, and the associated Fas, TNFR1, and F1A␣ were detected by Western blotting with anti-Myc antibody. These Myc-tagged proteins did not form nonspecific immunoprecipitates with the HA antibody (data not shown). The cytoplasmic domains of mFas-FD5, TNFR1-IC⌬15, and TNFR1-IC (Fig. 3B, lanes 2, 4, and 5) associated with F1A␣, whereas FADD, NGFR-IC, the cytoplasmic domain of the p75 NGF receptor that is a closely related member of the TNF receptor superfamily, and mFas-IC did not. The inability of mFas-IC to interact with F1A␣ in the mammalian cells and HB12 in yeast suggests that an in vivo mechanism may have restricted its accessibility to the protein. Similar to other signaling components serving members of the TNF/NGF receptor superfamily, F1A␣ was able to self-associate in vivo (Fig. 3B, lane 6).
Deletion Analysis of F1A␣-To delineate the functional domain of F1A␣ responsible for the apoptotic effect, MCF7 cells were transiently transfected with expression vectors encoding the HA-tagged F1A␣ or various deletion mutants. Interestingly, wild type F1A␣ exhibited slower kinetics in inducing apoptosis than FADD (Fig. 4A). When cells were scored for apoptosis 24 h post-transfection, F1A␣ appeared to be inactive, whereas FADD was apoptotic (Fig. 4A). However, when the transfected cells were examined at 36 and 48 h post-transfection, substantial apoptotic activity was demonstrated by F1A␣. At the 48-h time point, F1A␣ was as potent as FADD in inducing apoptosis in MCF7 upon overexpression (Fig. 4A). In our initial efforts to characterize the cDNA obtained from the yeast two-hybrid library, we observed a significant level of apoptosis in cells overexpressing HB12 at the 24-h time point, suggesting that HB12 was killing the cells with a different kinetics from F1A␣. Since HB12 is an N-terminal truncated form of F1A␣, it raised the possibility that the N-terminal region contains a negative regulatory domain. To facilitate the identification of the potential negative regulatory domain in F1A␣, we subjected F1A␣ and its deletion mutants to apoptosis assay at both 24-and 48-h time points. appeared to be substantially more effective in inducing apoptosis than HB12 at both time points, suggesting that additional ankyrin repeats other than the two found in HB12 are required for full apoptotic activity. The absence of apoptotic function in the C-terminal region was demonstrated by the overexpression of two C-terminal mutants, F1A␣-(343-627) and F1A␣-(482-627) (data not shown, Fig. 4B). Overexpression of F1A␣-(1-253) and F1A␣-(82-253) did not induce apoptosis (Fig. 4B, data not  shown). Thus the region immediately distal to the ankyrin repeat of F1A␣ is essential for apoptotic function, and the minimum effector domain is from amino acids 82 to 342. F1A␣-(1-530) and F1A␣-(1-342) exhibited apoptotic activity only at the 48-h time point, providing further support for the possible regulatory role of the N-terminal region of F1A␣. C-terminally tagged constructs gave identical results (data not shown). Western blot analyses showed that all the mutants were ex-pressed at the 24-h time point, and protein levels were generally higher among mutants that were not apoptotic (Fig. 4D). The presence of a negative regulatory domain that regulates the potency of a pro-apoptotic molecule has been suggested in Bim (44). Bim is a member of the "BH3 domain-only" family of pro-apoptotic proteins for which splice variants have been described. The three isoforms are very similar but there are clear differences in their cytotoxicity upon overexpression. Since the short form was the most potent inducer of cell death, the regions specific to Bim L and Bim EL were suggested to have a negative regulatory role (44).
To ensure that the apoptotic effect of F1A␣ is not restricted to MCF7 cells, the effects of transient expression of F1A␣ and its deletion mutants in HeLa and NIH3T3 cells were evaluated. In both cell types, F1A␣ overexpression resulted in cell death that could be blocked by treatment with 20 M Z-VAD-fmk or DEVD-fmk at the 48-h time point (Table I). Similar to that observed in MCF7 cells, the presence of the N-terminal region affected the kinetics of killing suggesting a negative regulatory role for the N-terminal region. The apoptotic activity of HB12 was also compromised in HeLa and NIH3T3 cells supporting the suggestion that a cluster of five ankyrin repeats is required for full apoptotic activity. caspase-3, -6, and -7 were able to cleave PARP and caspase-8 cleaved Bid (data not shown), only caspase-3 cleaved F1A␣ into a ϳ38and a ϳ32-kDa fragment (Fig. 5A). Several potential caspase cleavage sites were identified in F1A␣ by sequence inspection. Cleavage at two of these, i.e. DNID 342 and VYAD 356 , would generate fragments of the predicted sizes. To confirm the cleavage site, the aspartic acids of F1A␣ at position 342 and 356 were mutated to alanine. F1A␣(D342A) was resistant to caspase-3 cleavage, whereas F1A␣(D356A) was cleaved in a similar manner as the wild type F1A␣ (Fig. 5C,  lanes 5 and 6). The predicted sizes of the F1A␣ proteolytic fragments resulting from cleavage at Asp 342 are consistent with what is observed: 1-342 (38 kDa) and 343-627 (32 kDa) (Fig. 5B). The apoptotic N-terminal deletion mutant F1A␣-(82-627) was also susceptible to caspase-3 cleavage, yielding cleavage products of expected sizes, 82-342 (29 kDa) and 343-627 (32 kDa) (Fig. 5A, lane 7). We were unable to detect the cleavage products of F1A␣ in MCF7 and HeLa cells. Detection of cleavage product of caspases that is capable of inducing cell death has been shown to be technically challenging in transient transfection experiments (45).
Caspase inhibitors block the apoptotic effect of F1A␣, and this protein can be cleaved by caspase-3 in vitro, yielding F1A␣-(1-342) as one of the cleavage products. Furthermore, F1A␣-(1-342) was found to be apoptotically active by deletion analysis. These observations raise the possibility that proteolytic cleavage of F1A␣ may be required for its apoptotic function. To test this possibility, the wild type and cleavage-resistant forms of F1A␣, F1A␣(D342A), were overexpressed in MCF7 cells. The proteins were expressed at comparable levels as verified by Western blot analysis (data not shown). F1A␣(D342A) was inactive in the apoptosis assay (Fig. 5D), whereas F1A␣ and the control point mutant F1A␣(D356A), which could be cleaved by caspase-3 in vitro, were apoptotic. F1A␣(D342A) was able to associate with the death receptors such as mFas-FD5 and TNFR1-IC in vivo as indicated by immunoprecipitation assay (data not shown) suggesting that the overall conformation of the protein was still intact. Taken together these data suggest that proteolytic cleavage of F1A␣ at Asp 342 is a prerequisite for its apoptotic activity.
Apoptotic Effect of F1A␣ Is Blocked by Bcl-XL and Dominant Negative Mutants of FADD and Caspase-9 -Apoptosis mediated by death receptors (Fas/CD95 and TNFR1) involves FADD recruitment of caspase-8 and its subsequent proteolytic activation (23,25,46). However, in a variety of cell types, apoptotic signaling in response to Fas or TNFR1 activation is regulated at least in part by a Bcl-2 and/or Bcl-XL-inhibitable step (30,47). To establish a possible link between F1A␣ and components of various apoptotic pathways, blocking experiments using the anti-apoptotic Bcl-XL (48) and dominant negative mutants of several signaling molecules were performed. Dominant negative mutant of caspase-8, caspase-8-(1-415), which has been shown to block death receptor-mediated cell death (46), was only marginally effective in blocking apoptosis induced by overexpression of F1A␣ in MCF7 cells (Fig. 6). In contrast, Bcl-XL and dominant negative mutants of caspase-9, caspase-9-DN (49,50), and FADD, FADD-DN (51), were potent inhibitors of apoptosis induced by F1A␣ overexpression (Fig. 6) suggesting that they might have a role in the signaling pathway of F1A␣.

DISCUSSION
In this report, we describe the identification and characterization of F1A␣, a novel death receptor binding protein. F1A␣ does not have a death domain; however, it has six ankyrin motifs arranged in tandem at the N terminus. Ankyrin repeat structures are protein-protein interaction domains capable of associating with diverse proteins through heterotypic interactions (52,53). The ankyrin repeat motifs are found in F1A␣-(1-342), which apparently is important in mediating the apoptotic effect of F1A␣. However, F1A␣-(1-342) was unable to interact with the cytoplasmic domain of mFas or TNFR1 as suggested by co-immunoprecipitation experiment (data not shown). Deletion of 93 amino acid residues from the C terminus abolished the ability of F1A␣ to associate with these proteins. The 145-amino acid C-terminal region of F1A␣ alone is sufficient to interact with both the death receptors (data not shown) suggesting that the death receptor binding and death effector domains are separable in F1A␣ similar to that reported in the FADD protein (25). Therefore, F1A␣-(1-342) is likely to interact with another protein in the apoptotic pathway. The identification of this protein partner(s) would provide further in-  35 S-F1A␣-(82-627) (5 ϫ 10 5 cpm) were assayed for cleavage by bacterially expressed active forms of caspase-3 (0.1 g), caspase-6 (0.2 g), caspase-7 (0.1 g), and caspase-8 (0.5 g) as described under "Experimental Procedures." B, schematic representation of the putative caspase-3 cleavage sites in F1A␣ as estimated by the molecular mass of the cleavage products and the positions of the aspartic acid. Also shown is the sequence (one-letter code) between amino acids 335 and 360 where the aspartic acids in the tetrapeptides DNID (in bold type) and VYAD are replaced with alanine residues to yield F1A␣(D342A) and F1A␣(D356A) respectively. C, F1A␣ is cleaved by caspase-3 at Asp 342 . In vitro translated 35 S-labeled F1A␣, F1A␣(D342A), or F1A␣(D356A) was incubated with caspase-3, and the cleavage products were detected by autoradiography. D, cleavage-resistant mutant of F1A␣ failed to execute apoptosis upon overexpression. MCF7 cells were transfected with the indicated constructs and pCMV-␤-galactosidase followed by apoptosis assay. The data (mean Ϯ S.D.) shown are percentage of round blue cells as a function of total number of blue cells counted. sights into the molecular mechanism of F1A␣ action.
Caspase-3, but not caspase-6, -7, and -8, was found to cleave F1A␣ at a specific site, Asp 342 , situated at the C-terminal boundary of the minimum effector domain of death in F1A␣. The caspase that processed F1A␣, however, may not be caspase-3 because F1A␣ is not as susceptible as DFF45/ICAD, which is an established substrate of caspase-3 (54), to caspase-3 digestion in vitro. Whereas DFF45/ICAD was completely cleaved, only a fraction of F1A␣ was cleaved under the same experimental condition (data not shown). Therefore, a caspase-3-like rather than caspase-3 activity is likely to regulate the apoptotic activity of F1A␣. More than a dozen caspases have been identified in mammals (7); it is possible that caspases other than those used in the present study are able to process F1A␣ more efficiently. A single point mutation (D342A) in F1A␣ resulted in a cleavage-resistant mutant devoid of apoptotic activity suggesting the C-terminal domain of F1A␣ might also have a negative regulatory role. The activation of a pro-apoptotic molecule by caspase cleavage is well documented in BID, a "BH3 domain only" member of the Bcl-2 family. Upon activation of the TNFR1 or Fas apoptotic signaling pathways, BID is cleaved by caspase-8 to generate a C-terminal fragment, tBID, which is a potent inducer of cytochrome c release and apoptosis. Cleavage of BID by caspase-8 therefore relieves the inhibitory effect of the N-terminal region that controls its proapoptotic activity (10,11,13,63).
Overexpression of F1A␣ induced apoptosis in MCF7 cells and co-expression of Bcl-XL or the dominant negative mutants of either FADD or caspase-9 could diminish this apoptotic effect. These observations suggest that F1A␣ may be a component of a signaling pathway that involves FADD, caspase-9, and Bcl-XL. The dominant negative mutant of FADD may compete with F1A␣ for binding to the death receptor, thus diminishing its apoptotic activity. Fas signaling is thought to diverge at caspase-8 with one branch of the pathway leading directly to effector caspase activation and the other branch communicating with the mitochondria that are caspase-9-and Bcl-XL-dependent (10,11,30). In MCF7 cells because of caspase-3 deficiency, the predominant pathway for Fas signaling is Bcl-XLdependent (30). The blocking data thus suggest that F1A␣ most probably play a role in the Fas signaling pathway upstream of Bcl-XL and caspase-9.
F1A␣ shares substantial amino acid sequence homology (ϳ30% identity) with the gene product of the C. elegans sexdetermining gene, fem-1. The fem-1 gene was identified in genetic screens and is required for sex determination of male phenotype in both germ line and somatic tissues in the nematode C. elegans (55,56). We have demonstrated the ability of F1A␣ to induce apoptosis, which raises the question about FEM-1 function in C. elegans. The current understanding of sex determination in C. elegans during development does not appear to rule out an apoptotic role for FEM-1. In C. elegans, the earliest sex-specific events that occur during embryogenesis are two sets of cell deaths, one male and the other hermaphrodite-specific (57). In the male embryo, two motor neurons called HSNs undergo apoptosis at hour 8 of embryonic development (hatching occurs at hour 13). At about the same time in the hermaphrodite embryo, four sensory neurons called CENs also die by apoptosis. The molecular pathways that control these sex-specific apoptosis events are not well understood.
Despite its exclusive role in masculinizing somatic tissues in males and regulating the production of male germ cells in both males and hermaphrodites, FEM-1 protein is expressed throughout development in all somatic tissues at equivalent levels in both sexes (58). The activity of FEM-1 is therefore thought to be controlled post-translationally. Alternatively, FEM-1 may have a role other than sex determination in C. elegans.
The relatively high degree of sequence homology between FEM-1 and F1A␣ is intriguing considering that two other genes flanking fem-1 in the sex determination pathway, tra-1 and tra-2, are the most highly diverged genes compared between the two Caenorhabditis species, C. elegans and Caenorhabditis briggsae (59,60). Although the sex determination pathways in mammals and C. elegans are thought to be quite different, recent evidence suggests that certain regulatory proteins in these pathways are indeed conserved (61,62). Whether F1A␣ has a role in determining the sexual fate in mammals requires further investigation.