dFADD, a Novel Death Domain-containing Adapter Protein for theDrosophila Caspase DREDD*

Apoptotic cell death occurs through activation of procaspases, the precursors of a group of aspartate-specific cysteine proteases known as caspases. Procaspase activation is mediated by death adapter proteins such as the mammalian proteins FADD and Apaf-1 and theCaenorhabditis elegans protein CED-4. These adapters bind to procaspases and facilitate oligomerization and subsequent auto-proteolytic processing of the zymogens. Here we report cloning and characterization of dFADD, a FADD homologue inDrosophila. dFADD contains a death domain that is highly homologous to the FADD death domain, and it also shares a novel domain with a Drosophila caspase DREDD, which we call death-inducing domain. dFADD binds to DREDD through the death-inducing domain and enhances the cell death activity and proteolytic processing of DREDD. dFADD and DREDD are stabilized by their interaction. The structural and functional similarities between dFADD and FADD suggest the existence of a FADD-like apoptosis pathway in Drosophila.

by cell-intrinsic developmental cues or cytotoxic reagents and is found in the nematode C. elegans and mammals (9,10). Upon apoptosis activation, the death adapter proteins CED-4 and Apaf-1 form homo-oligomers that subsequently aggregate the CED-3 caspase and procaspase-9, respectively (11)(12)(13), through homotypic interactions mediated by the caspase recruitment domain (CARD) 1 (14). Mammals have also evolved an extrinsic or instructive apoptosis pathway, mediated by small death adapters such as FADD, that allows a cell to instruct another to undergo self-destruction (15,16). This pathway is engaged by a unique group of death receptors in the tumor necrosis factor receptor superfamily (e.g. TNFR1 and Fas) and plays important roles in the regulation of immune responses and the maintenance of homeostasis. Upon binding to their trimeric ligands, these receptors recruit the adapter protein FADD via death domain (DD)-DD interaction (6). FADD then binds to procaspase-8 through another homotypic interaction involving death effector domain (DED), a motif present in the N-terminal region of both proteins (17,18). Recruitment of procaspase-8 to the ligand-aggregated death receptors leads to its oligomerization and subsequent activation (19 -21).

MATERIALS AND METHODS
Cell Lines, Expression Vectors, and Reagents-Human embryonic kidney 293 cells, human cervical carcinoma HeLa cells, and murine fibroblast 3T3 cells were cultured in complete Dulbecco's modified Eagle's medium, and human breast carcinoma MCF7 cells were maintain in RPMI 1640. Expression constructs were based on pRK5 (a gift from Dr. D. Goeddel), pcDNA3 (Invitrogen), pEGFP-C1 (CLONTECH; for Aequorea victoria green fluorescence protein (GFP) fusion constructs), and pCaSpeR-hs (a gift of Dr. M. Fortini; for SL2 cell expression constructs). FLAG and HA-tags were placed at the N termini of the fusion proteins. Anti-FLAG and anti-HA polyclonal antibodies (Santa Cruz) and anti-FLAG mAb M2 conjugated on agarose beads (Sigma) were purchased from the indicated sources.
Cloning of dFADD-A Drosophila expressed sequence tag clone (GenBank TM accession number AI294992) was found to contain an open reading frame with significant sequence homology to mammalian death domain-containing proteins. The sequence of the open reading frame was confirmed on both strands with an automated sequencer (Applied Biosystems).
Northern Blotting-A Northern blot membrane containing mRNA from different developmental stages of Drosophila embryos was hybridized with a 32 P-labeled cDNA probe corresponding to the DNA sequence of the prodomain (residue 1-141) of dFADD or a probe specific for ribosomal RNA RP49 as a loading control.
Transfection, Coimmunoprecipitation, and Western Analysis-Transient transfection of 293 cells was performed as described previously (25). Unless indicated otherwise, cells were transfected with 1-10 g of indicated plasmids to yield equivalent protein expression levels. 20 -24 * This work was supported by an academic development fund from the University of Pennsylvania (to X. Y.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM  h after transfection, cell lysates were prepared and immunoprecipitated with anti-FLAG mAb M2 beads. The precipitates were resolved by SDS polyacrylamide gel electrophoresis and analyzed by immunoblotting with polyclonal anti-FLAG and anti-HA antibodies.
Cell Death Assay-HeLa, 3T3, and MCF7 cells were transiently transfected with each of the test plasmids plus the reporter plasmid pCMV-lacZ (1 and 0.2 g, respectively, unless indicated otherwise). 22-24 h after transfection, cells were fixed in 0.5% glutaraldehyde and stained with X-Gal. The percentage of apoptotic cells was determined by the number of membrane-blebbing cells divided by the total number of blue cells. Data presented were obtained from representative experiments performed in duplicates, and the mean and standard deviation were calculated.
Drosophila Schneider L2 cells were transfected with dFADD and DREDD expression plasmids together with a GFP expression plasmid. 20 h after transfection, cells were heat shocked for 20 min three times at 37°C. 24 h after the heat shock, cells were examined under a fluorescence microscope to determine the number and morphology of GFP-positive cells.
Fluorescence Microscopy-HeLa cells were transfected with the GFP fusion plasmids. 20 -24 h later, the cells were examined under a fluorescence microscope to determine the distribution and the relative intensity of fluorescence.

RESULTS AND DISCUSSION
Identification of a Drosophila Death Domain-containing Protein, dFADD-In a search for Drosophila death domain-containing proteins in the expressed sequence tag data base, we identified a full-length cDNA predicted to encode a protein with a death domain at its C terminus (Fig. 1). This protein has an overall structure similar to that of FADD, and its death domain is highly homologous to the FADD death domain (28% identity FIG. 2. dFADD specifically associates with DREDD through DID-DID interaction. A, schematic diagram of expression constructs for dFADD, DREDD, and their mutants. Amino acids at the boundaries of DID, DD, caspase domain (casp), and deletion mutants are labeled. p15 and p10, the large and small subunits of the DREDD caspase domain, respectively. The dFADD and DREDD constructs contained N-terminal HA and FLAG tags, respectively. B, dFADD binds DREDD but not DRONC. 293 cells were co-transfected with FLAG-tagged DRONC (FLAG-DRONC) or FLAG-DREDD together with vector or HA-tagged dFADD (HA-dFADD) as indicated. The amount of the plasmids was adjusted to yield approximately equal levels of protein expression among different transfections. 20 -24 h after transfection, cells lysates were prepared and immunoprecipitated with anti-FLAG mAb M2 conjugated on agarose beads. The precipitates (IP) and extracts (Ext) were analyzed by SDS polyacrylamide gel electrophoresis and immunoblotting (WB) with polyclonal anti-HA or anti-FLAG antibody as indicated. Upon overexpression, FLAG-DREDD underwent auto-processing to generate DREDD-⌬p10 (see Fig. 4A). C, interaction of dFADD with various caspases. 293 cells were transfected with HA-dFADD plus indicated FLAG-tagged mammalian caspases. The anti-FLAG immunoprecipitates and extracts were analyzed as in B. In the bottom panel, FLAG-caspase bands from the same immunoblot were aligned to show equivalent protein levels. D, dFADD interacts with DREDD through the DID. E, DREDD binds to dFADD through the prodomain. F, the DID domain of DREDD is required for dFADD association. In D, E, and F, 293 cells were transfected with indicated plasmids, and co-immunoprecipitation assay was performed as in B. and 49% similarity) (Fig. 1B). We hence designated this first Drosophila death domain-containing protein dFADD.
No homology was found between the N-terminal region of dFADD and any other proteins in the data base using the BLAST and SMART programs. However, a careful comparison of the dFADD sequence with the sequence of DREDD, a Drosophila apical caspase, revealed significant homology between the dFADD N-terminal region and a region in the DREDD prodomain (19% identity and 39% similarity) (Fig. 1C). Although DREDD was thought to contain two DED domains (26), we found no homology between DREDD prodomain and any DED-containing proteins using BLAST or SMART. The conserved domain in dFADD and DREDD is distinct from DED and CARD (two domains involved in death adapter-procaspase interaction), but its function is similar to those of DED and CARD (see below). We therefore named this novel domain DID (death-inducing domain). Analysis of the dFADD genomic sequence revealed that it contains a single exon on chromosome 3.
Expression of dFADD mRNA-To examine the mRNA expression of dFADD, we hybridized mRNAs sampled from different stages of Drosophila embryonic development with a dFADD cDNA probe. Only the 3-to 12-h embryos contained a single 1.7-kilobase dFADD transcript (Fig. 1D). Thus, the expression of dFADD is tightly regulated during Drosophila development.
dFADD Specifically Associates with DREDD-Because death adapter proteins associate with caspases through homotypic DED-DED or CARD-CARD interaction, we investigated whether dFADD interacted with DREDD through their homologous DIDs. Co-immunoprecipitation assays confirmed that dFADD specifically associated with DREDD but not with DRONC, a CARD domain-containing Drosophila apical caspase (27) (Fig. 2B). The specificity of the dFADD-DREDD interaction was further underlined by the observation that dFADD did not interact with various mammalian caspases except for its weak interaction with caspase-10 (Fig. 2C). The interaction domains in dFADD and DREDD were mapped using a panel of deletion mutants (Fig. 2A). The N-terminal region of dFADD containing the DID domain was both necessary and sufficient for the interaction with DREDD (Fig. 2D). dFADD interacted strongly with the DREDD prodomain but

. Enhancement of DREDD processing by dFADD and comparison of dFADD-DREDD and DARK-DREDD interactions.
A, dFADD enhances DREDD processing. 293 cells were transfected with 3 g of DREDD or DRONC and 5 g of dFADD together with p35 and CrmA. Cell extracts were immunoprecipitated with anti-FLAG mAb M2 antibody and analyzed by immunoblotting with polyclonal anti-FLAG antibody. B, self-association of DREDD. 293 cells were transfected with indicated plasmids, and co-immunoprecipitation assay was performed as in A. DREDD-C/S, the DREDD active site cysteine to serine mutation. Asterisks, nonspecific bands serving as protein-loading control. C and D, DREDD interacts strongly with dFADD but weakly with DARK. Transfection and co-immunoprecipitation assays were performed as described in Fig. 2B. weakly with the caspase domain (Fig. 2E). Further mutagenesis analysis of the DREDD prodomain region revealed that the DID domain was required for the interaction with dFADD (Fig.  2F). Therefore, the DID domains mediate the interaction between DREDD and dFADD.
dFADD Enhances DREDD-induced Apoptosis-To examine the effect of dFADD on DREDD-induced apoptosis, we transfected Drosophila SL2 cells with dFADD and DREDD expression plasmids either alone or together. No apoptosis was observed (data not shown), consistent with previous results that ectopic expression of DREDD alone in SL2 cells did not lead to activation of the caspase (26). However, expression of DREDD in several mammalian cell lines caused significant apoptosis (Fig. 3, A, B, and C). We examined the effect of dFADD on DREDD-mediated apoptosis in these cells and found that although dFADD did not induce apoptosis, it potently enhanced DREDD-mediated apoptosis in a dose-dependent manner (Fig.  3, A, B, and C). The dFADD DID was both necessary and sufficient for this enhancement. dFADD did not enhance DRONC-induced apoptosis, consistent with the lack of interaction between these two proteins (Fig. 3D).
dFADD Promotes DREDD Processing-Initiator caspases are activated by adapter-mediated oligomerization and subsequent auto-processing. Because dFADD associated with DREDD and enhanced DREDD-induced apoptosis, we investigated whether dFADD promoted DREDD processing. Overexpression of DREDD in human 293 cells led to zymogen processing (Fig. 4A). DREDD associated with itself at high expression levels either directly or indirectly via an endogenous adapter protein (Fig. 4B), and such self-association could facilitate its auto-cleavage. The DREDD processing was not inhibited by p35 and crmA, two active site-specific caspase inhibitors that can inhibit most caspases ( Fig. 4A and data not shown), and DREDD may possess unique substrate specificity. In the presence of dFADD, DREDD processing went to completion (Fig. 4A), whereas DRONC was not processed upon overexpression either in the presence or absence of dFADD (Fig. 4A). It appears that the enhancement of DREDD processing by dFADD is mediated by direct protein-protein interaction.
Comparison of dFADD-DREDD and DARK-DREDD Interactions-DREDD was previously reported to associate with the Drosophila Apaf-1 homologue DARK (23). Other Drosophila caspases such as DRONC were also reported to be targets for DARK (22,24). We compared the dFADD-DREDD and DARK-DREDD interactions using co-immunoprecipitation assay and found that the former was much stronger than the latter. Unlike the dFADD-DREDD interaction that was mainly mediated by the DREDD prodomain (Fig. 2E), both the DREDD prodomain and the caspase domain interacted weakly with DARK (Fig. 4D). The strong dFADD-DREDD interaction suggests that DREDD may mainly serve as an apical caspase for dFADD.
Cellular Localization of dFADD and Stabilization of dFADD Protein by DREDD-To examine the cellular localization of dFADD and DREDD, we fused them to GFP. When expressed in HeLa cells, both GFP-dFADD and GFP-DREDD proteins were localized in the cytoplasm (Fig. 5A, a and c). This localization of dFADD was mediated by its N-terminal region (Fig.  5A, e and f). dFADD and DREDD appeared to stabilize each other, because expression of DREDD significantly increased fluorescence intensity in the GFP-dFADD-transfected cells (Fig. 5A, b versus a), and similarly, expression of dFADD enhanced fluorescence intensity in the GFP-DREDD-transfected cells (Fig. 5A, d versus c). We also compared the expression levels of dFADD in 293 cells in the presence and absence of DREDD. Co-transfection of dFADD with DREDD led to a higher level of dFADD in both the soluble and insoluble fractions of the cell extracts compared with transfection of FADD alone (Fig. 5B). In contrast, DRONC did not enhance dFADD protein expression (Fig. 5B). The stabilization of dFADD by DREDD was mediated by the DREDD prodomain ( Fig. 5C and data not shown). Reciprocally, dFADD stabilized DREDD but not DRONC in 293 cells (data not shown). The mutual stabilization of dFADD and DREDD suggests that the expression of these two proteins may be co-regulated in vivo.
In summary, we describe here the molecular cloning and partial characterization of the first Drosophila death domaincontaining protein dFADD. dFADD contains a bipartite structure highly homologous to that of the mammalian adapter FADD. It functions similarly to FADD and physically interacts with and activates the Drosophila caspase DREDD. The interaction between dFADD and DREDD stabilizes both proteins. We conclude that Drosophila also contains a FADD-like apoptotic pathway. To date, no death receptors have been identified in Drosophila, and despite the high homology between the death domains of FADD and dFADD, we did not detect interaction between dFADD and any of the mammalian death receptors (data not shown). Identification of components upstream of dFADD should help determine the regulation and function of this evolutionarily conserved apoptosis pathway.