Molecular Cloning and Characterization of DEFCAP-L and -S, Two Isoforms of a Novel Member of the Mammalian Ced-4 Family of Apoptosis Proteins*

We report the deduced amino acid sequences of two alternately spliced isoforms, designated DEFCAP-L and -S, that differ in 44 amino acids and encode a novel member of the mammalian Ced-4 family of apoptosis proteins. Similar to the other mammalian Ced-4 proteins (Apaf-1 and Nod1), DEFCAP contains a caspase recruitment domain (CARD) and a putative nucleotide binding domain, signified by a consensus Walker's A box (P-loop) and B box (Mg2+-binding site). Like Nod1, but different from Apaf-1, DEFCAP contains a putative regulatory domain containing multiple leucine-rich repeats (LRR). However, a distinguishing feature of the primary sequence of DEFCAP is that DEFCAP contains at its NH2 terminus a pyrin-like motif and a proline-rich sequence, possibly involved in protein-protein interactions with Src homology domain 3-containing proteins. By using in vitro coimmunoprecipitation experiments, both long and short isoforms were capable of strongly interacting with caspase-2 and exhibited a weaker interaction with caspase-9. Transient overexpression of full-length DEFCAP-L, but not DEFCAP-S, in breast adenocarcinoma cells MCF7 resulted in significant levels of apoptosis. In vitro death assays with transient overexpression of deletion constructs of both isoforms using β-galactosidase as a reporter gene in MCF7 cells suggest the following: 1) the nucleotide binding domain may act as a negative regulator of the killing activity of DEFCAP; 2) the LRR/CARD represents a putative constitutively active inducer of apoptosis; 3) the killing activity of LRR/CARD is inhibitable by benzyloxycarbonyl-Val-Ala-Asp (OMe)-fluoromethyl ketone and to a lesser extent by Asp-Glu-Val-Asp (OMe)-fluoromethyl ketone; and 4) the CARD is critical for killing activity of DEFCAP. These results suggest that DEFCAP is a novel member of the mammalian Ced-4 family of proteins capable of inducing apoptosis, and understanding its regulation may elucidate the complex nature of the mammalian apoptosis-promoting machinery.

Apoptosis (programmed cell death) is the genetically determined cell suicide program resulting in distinct biochemical and morphological features. Some of the hallmark characteristics of apoptosis include plasma membrane blebbing, nuclear and cytosolic condensation, and ultimately the formation of membrane-bound apoptotic bodies primed for phagocytosis (1). Alterations in the ability of the cell to initiate and/or execute the proper apoptotic signaling cascade have been implicated in many diseases such as cancer, autoimmune diseases, viral infections, and neurodegenerative disorders (2). Identifying the key mediators of apoptosis and understanding the molecular mechanisms of programmed cell death is critical to understanding the pathogenesis of these diseases.
Genetic studies conducted by Horvitz and co-workers (3, 4) using the roundworm, Caenorhabditis elegans, identified the genes ced-3, ced-4, and ced-9 as crucial for initiating proper apoptosis in the developing nematode. Loss of function mutations in Ced-3 and Ced-4 or an overexpression of Ced-9 prevented programmed cell death in the developing nematode. The identification and characterization of these genes led to a wealth of research into the identification of their mammalian counterparts.
Based on the analysis of Ced-3 homologous sequences, the cysteine protease interleukin 1␤-converting enzyme was identified as the first mammalian Ced-3-like protein, which led to the identification of a family of proteases named caspases (cysteine protease with cleavage specificity after an aspartate residue). Caspases exist in the cell as inactive zymogens or procaspase molecules that with the proper stimuli (e.g. death receptor ligation, DNA damage, and chemotherapeutic agents) can be recruited via their caspase recruitment domain (CARD) 1 to form homo-and heterodimers. Following their recruitment, caspases can either autocatalytically or via another caspase become processed into two catalytic subunits and subsequently converted into active enzymes that are capable of initiating the execution arm of the apoptotic signaling pathway. To date, at least 14 members of the caspase family have been identified. Proapoptotic caspases can be classified into two classes based on their primary sequence. Upstream regulatory caspases in-* 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 /EBI Data Bank with accession number(s) NP_055737.
To date, only two mammalian Ced-4 homologues, Apaf-1 (5) and Nod1 (6) CARD4 (7), and one Drosophila Ced-4 homologue, DAPAF-1 (8)/DARK (9), have been reported. Apaf-1, Nod1, DAPAF-1, and Ced-4 are all similar in having an NH 2 -terminal CARD followed directly by a nucleotide binding domain (NBD), also known as the NB-ARC or NOD domain. Apaf-1 and Nod1 differ from Ced-4 by having a COOH-terminal regulatory domain containing multiple WD-40 repeats for Apaf-1 and multiple leucine-rich repeats (LRR) for Nod1. After the initiation of an apoptotic stimulus and release of cytochrome c from the mitochondria, cytochrome c is believed to bind the WD-40 repeats of Apaf-1 causing a conformational change in Apaf-1. Post-translational modifications (i.e. cytochrome c binding) of Apaf-1 can lead to Apaf-1 oligomerization and the subsequent ATP-dependent recruitment, processing, and activation of caspase-9 (5, 10, 11). Apaf-1, caspase-9, and other core components of the apoptosis machinery interact to form an ϳ700-kDa biologically active and ϳ1.4-MDa biologically inactive complex known as the apoptosome (12). Formation of the apoptosome and the subsequent processing of caspase-9 have been shown to be crucial for the activation of effector caspases, such as caspase-3, leading to the demise of the cell. In this model, Apaf-1 serves as an adaptor molecule linking upstream regulatory caspases with downstream effector caspases.
Nod1, on the other hand, appears to function in activating the NF-B pathway via an interaction with RICK, a CARDcontaining kinase. Furthermore, Inohara et al. (6) demonstrate that Nod1 is capable of interacting with caspase-9 and enhancing caspase-9-mediated apoptosis. Nod1/RICK-induced NF-B activation and Nod1/caspase-9-mediated apoptosis are believed to occur via independent pathways based on the results that an active site caspase-9 mutant was not able to inhibit Nod1mediated NF-B activation (13).
Here we report the deduced amino acid sequences and characterization of DEFCAP-L and DEFCAP-S (Death Effector Filament-forming Ced-4-like Apoptosis Protein), two isoforms for a novel member of the mammalian Ced-4 family of proteins. Named for the ability of CARD to form novel cytoplasmic structures termed death effector filaments (14), DEFCAP-L and DEFCAP-S differ in only 44 amino acids, containing an extra LRR. The NH 2 -terminal region of DEFCAP consists of a pyrinlike motif (PLM), a proline-rich sequence (PR), followed by a highly conserved NBD, a multiple LRR containing domain, and a COOH-terminal CARD. Based on the primary sequence of DEFCAP, which contains a well conserved CARD, NBD, and a putative regulatory domain with multiple LRR motifs, we classify DEFCAP as the third member of the mammalian Ced-4 family of proteins.

MATERIALS AND METHODS
High fidelity polymerase chain reaction (PCR) reagents for cloning were purchased from Roche Molecular Biochemicals. DNA sequencing was performed by the University of Michigan Core Facilities, and oligonucleotide synthesis was performed by Operon Technologies, Inc. Restriction enzymes, linkers, and other modifying enzymes were purchased from New England Biolabs. Antibodies to c-Myc (9E10) AC, caspase-8 p20 (C-20), caspase-9 p46 (H-170), and caspase-10 (H- 19) were purchased from Santa Cruz Biochemicals. Caspase-2 monoclonal antibodies (I75620) were purchased from PharMingen/Transduction Laboratories. RAIDD monoclonal antibodies (4B12) were purchased from StressGen Biotechnologies Corp. ZVAD-fmk was purchased from Enzyme Systems Products.
Mammalian expression vector pcDNA3.1 was purchased from In-vitrogen and modified to contain an NH 2 -terminal Myc epitope (pcDNA3.1 NmycH) by annealing sense and antisense oligonucleotides coding for the Myc epitope followed by digestion and ligation into the NheI and HindIII sites. 5Ј-Sequences flanking the Myc epitope were converted to Kozak consensus translation initiation sites. Northern Analysis and Semi-quantitative RT-PCR-Human 12-lane MTN blot (catalog number 7780-1) and a human cancer cell line MTN blot (catalog number 7757-1) were obtained from CLONTECH. Fulllength DEFCAP-L cDNA was radiolabeled with [␣-32 P]dCTP, hybridized, and washed according to the protocol described previously (15) and exposed to autoradiography film for 5 days. ␤-Actin cDNA was used as a control for equal loading of RNA. Random primed reverse transcription for semi-quantitative RT-PCR was performed using Superscript II reverse transcriptase (Life Technologies, Inc.) according to the manufacturer's protocol. RNAs from normal human liver, spleen, polymorphonuclear cells (PMNs), peripheral blood mononuclear cells (PBMCs), K562 and Jurkat cancer cell lines, and a negative control without template were reversed-transcribed. RNAs from K562 and Jurkat cancer cell lines were isolated using the Trizol reagent (Life Technologies, Inc.)and PMN and PBMC RNAs were isolated by a Ficoll-Paque (Amersham Pharmacia Biotech) gradient, following dextran sedimentation and hypotonic red blood cell lysis. 1 l of the RT reaction was used for PCR with the following oligonucleotides: 5Ј-CGAGAACAGCTGGTCT-TCTCCAGGGCTTCG (antisense) and 5Ј-TCCCCCTTGGGAGTCCTC-CTGAAAATGATC (sense) under the following conditions: 1 cycle at 94°C for 5 min, 30 cycles at 94°C for 30 s, 65°C for 30 s, 72°C for 30 s, and 1 cycle at 72°C for 5 min. ␤-Actin oligonucleotides 5Ј-CGAGAA-GATGACCCAGATCATGTTTGAGAC (sense) and 5Ј-TGGAAGCAGC CGTGGCCATCTCTTGCTCGA (antisense) were used as a control for RNA integrity and RT efficiency. PCR products were separated by agarose gel electrophoresis, stained with ethidium bromide, and photographed under UV light.
Chromosomal Localization-The Stanford G3 Human/Hamster Radiation Hybrid Panel (Research Genetics) was screened by PCR with the oligonucleotides 5Ј-CAGCGTCTCCAAGCTCAGCCATTGGGA-CCC-3Ј (sense) and 5Ј-CAGGCCATGTATTCCATATGCTTCTAGCGT (antisense) yielding an amplified product of ϳ300 bp encoding amino acids 560 -665. These primers were believed to be restricted to a single exon based on the identification of continuous genomic sequences from PAC pDJ891a18 identified by a nucleotide search of the High Through put Genomic Sequences database.
Construction of EGFP Constructs and Fluorescence Microscopy-The multiple cloning site of pEGFPC1 (CLONTECH) was digested with BamHI, Klenow filled-in to create blunt ends, and ligated with NotI linkers (New England Biolabs) to create pEGFPC1-NotI. The NH 2terminal EGFP DEFCAP-CARD (a.a. 1356 -1473) fusion construct was made in pEGFPC1-NotI by ligating DEFCAP sequences in frame using 5Ј KpnI and 3Ј NotI restriction sites. Cells were visualized at ϫ 100 magnification by fluorescence microscopy using a PixCell II microscope (Arcturus).
Cell Culture and Transient Transfection-Human embryonic kidney 293 cells were maintained in 10% fetal calf serum with Dulbecco's modified Eagle's medium supplemented with L-glutamine, penicillin/ streptomycin, and nonessential amino acids. K562 human erythroleukemia cells and MCF7, a human breast carcinoma cell line, were grown in 10% heat-inactivated fetal calf serum in RPMI 1640 with L-glutamine, penicillin/streptomycin, and nonessential amino acids. 293 cells were transfected using standard CaPO 4 Ϫ precipitation, and MCF7 cells were transiently transfected using LipofectAMINE (Life Technologies, Inc.) according to the manufacturer's protocol.
Immunoprecipitation and Western Blotting-293 cells were seeded at 10% confluency on 10-cm dishes, grown to 50% confluency, and prior to transfection were given fresh media. A total of 10 g of DNA was transfected into each dish as stated previously. 6 h post-transfection, the transfected cells were given fresh media, and cells were harvested 12-24 h post-transfection. Cells were harvested by collecting and pooling floating cells with adherent cells on ice and washed once with ice-cold phosphate-buffered saline (PBS). The cells were resuspended in 1 ml of lysis buffer containing 1% Nonidet P-40, 20 mM Tris, pH 8.0, 137 mM NaCl, 10% glycerol, 2 mM EDTA, 5 mM sodium vanadate, 0.5 mM phenylmethylsulfonyl fluoride, and 1ϫ protease inhibitors (complete, Mini, BMB 1 836 153) and incubated on ice for 15 min. The cell debris was spun down at 20,000 ϫ g for 15 min, and the supernatant was transferred to a fresh tube. The pellet was washed once with ice-cold PBS, mixed with 25 l of 1.5ϫ SDS-polyacrylamide gel electrophoresis sample buffer containing dithiothreitol, boiled at 100°C for 5 min, and saved at Ϫ20°C for Western analysis. The supernatant was precleared with protein G beads (Sigma) overnight at 4°C with gentle rotation. The following day, the protein G beads were spun down at 2000 ϫ g, and the supernatant was transferred to a new tube. 35 l of precleared supernatant was analyzed for Western analysis, and the remaining supernatant was used for immunoprecipitations conducted with gentle rotation at 4°C with the appropriate antibody for 3 h. DEFCAP constructs were immunoprecipitated with 10 l of Myc-AC. Following incubation, the immunoprecipitations were washed 4 times with PBS on ice, resuspended in 1.5ϫ SDS-polyacrylamide gel electrophoresis sample buffer, boiled, and analyzed by Western blotting.
Caspase Interaction Experiments-Coimmunoprecipitation experiments with either full-length NH 2 -terminal Myc-tagged DEFCAP-L or -S and caspases-2, -3, -8, -9, and -10 were performed in 293 cells. Briefly, 293 cells were cotransfected with 5 g of Myc-DEFCAP-L or Myc-DEFCAP-S and with 3 g of caspase construct. 18 -24 h post-transfection, cells were harvested and immunoprecipitated with Myc-AC and immunoblotted with caspase specific antibodies. The caspase-2 and caspase-9 coimmunoprecipitation experiments with both DEFCAP isoforms were reproduced in three independent experiments.
Cell Death Assays-MCF7 cells were plated on 35-mm 6-well tissue culture dishes, and each well was cotransfected at ϳ60% confluency with 0.25 g of the reporter plasmid pCMV ␤-galactosidase and 1 g of either pcDNA3.1 alone or an NH 2 -terminal Myc-tagged DEFCAP construct. 24 h post-transfection, the cells were fixed in 0.5% glutaraldehyde and stained with 5-bromo-4-chloro-3-indolyl ␤-D-galactopyranoside (X-gal) for 4 h. The cells were visualized by phase contrast microscopy, and the percentage of apoptotic cells was determined by counting at least 600 blue cells (n Ն 3). Round blue cells and/or blue cells exhibiting plasma membrane blebbing and cell shrinkage were scored apoptotic. The data presented were from at least two independent experiments conducted in duplicate or triplicate. A Student's t test using the computer program SigmaStat (Jandel) comparing vector control with various DEFCAP constructs was performed to obtain p values.

RESULTS AND DISCUSSION
Cloning of DEFCAP-L from a K562 cDNA Library and Chromosomal Assignment-By homology search using BLASTP for CARD-containing proteins, we identified clone KIAA0926 (GenBank TM accession number NP_055737), which was kindly provided by the Kazusa DNA Research Institute (21). For simplicity, clone KIAA0926 will be referred to as DEFCAP-S. By using the DEFCAP-S nucleotide sequence, we generated oligonucleotides that were used to PCR-amplify the DEFCAP open reading frame from a K562 cDNA library. PCR-amplified products were cloned and restriction-mapped, and DNA sequencing was performed on both strands to confirm the sequence accuracy of the clones. DNA sequences with translation products of perfect identity to DEFCAP-S were obtained from a K562 human erythroleukemia cancer cell line cDNA library. In addition, we identified some clones that contained a 132-bp insertion near the 3Ј end of the DEFCAP open reading frame. DNA sequencing followed by a BLAST search of the nonredundant data base identified a perfect match for human hypothetical protein DKFZp58601822.1 (GenBank TM accession number T17255). Analysis of the 132-bp insertion revealed that these sequences encode for an extra LRR not found in DEFCAP-S. We designated the full-length DEFCAP sequence containing the additional 132-bp sequence as DEFCAP-L.
By using the Stanford G3 radiation hybrid panel, we assigned the DEFCAP chromosomal localization between sequence-tagged sites (STSs) D17S849 and D17S796 of chromosome 17p13 (LOD score 7.53-9.25). These results are in agreement with those obtained by Nagase et al. using the Genebridge4 radiation hybrid panel for clone KIAA0926 (21).
The Amino Acid Sequences of DEFCAP-L and DEFCAP-S Share Homology with Apaf-1, Nod1, and Ced-4 -The deduced amino acid sequences for DEFCAP-L and DEFCAP-S encode proteins of 1473 and 1429 amino acids, respectively. Similar to the other mammalian Ced-4 homologues identified thus far, DEFCAP contains a CARD domain, a putative nucleotide binding domain (NBD), and a putative regulatory domain containing multiple repeat elements (LRRs). However, unlike the two other mammalian Ced-4 homologues Apaf-1 and Nod1, the positioning of these protein domains in the primary sequence is not conserved. Both Apaf-1 and Nod1 contain NH 2 -terminal CARD domains followed directly by an NBD. The COOH terminus of Apaf-1 is composed of 12-13 WD-40 repeats due to alternate splicing, whereas that of Nod1 is composed of 10 LRRs. Without knowing the crystal structure of the full-length DEFCAP protein or that of any other Ced-4 family member, it is difficult to comment on the significance of the COOH-terminal CARD found in DEFCAP versus the NH 2 -terminal CARD found in all other Ced-4-like proteins. However, the juxtaposition of the LRRs and the CARD in DEFCAP may result in an overall structure that is unique among the Ced-4 family members.
Some features of DEFCAP's primary sequence that distinguish it from all other Ced-4-like molecules are its NH 2 -terminal pyrin-like motif (PLM) (a.a. 1-95), reverse-highlighted in gray, and its proline-rich sequence (PR) (a.a. 40 -257) containing 9 PXXP motifs, underlined in black (Fig. 1A). DEFCAP's PLM shares 25% identity to pyrin or marenostrin, a CARDcontaining protein originally identified by positional cloning experiments in patients with Familial Mediterranean Fever disease, an inherited disease characterized by excessive neutrophil activity resulting in recurrent episodes of inflammation involving serosal and synovial spaces. The pyrin-like motif is conserved with other mammalian proteins such as ASC (apoptosis-associated speck-like protein containing a CARD), a COOH-terminal CARD-containing protein with an NH 2 -terminal PLM (22), which shares 28% identity with DEFCAP-PLM. Furthermore, the PLM is evolutionarily conserved as seen by a protein alignment with Danio rerio ASC1 (Fig. 1B). The recent identification of an emerging number of PLM and CARD-containing proteins suggests that the PLM may play a role in regulating the apoptotic machinery.
Directly following the PLM and PR of DEFCAP is a highly conserved Ced-4 homology domain or NBD (a.a. 309 -648) containing a consensus A box (P-loop), B box (Mg 2ϩ binding), and motif III, a conserved sequence with unknown function (all highlighted in red, Fig. 1A). Asterisks below the residues denote the conserved amino acids as determined by Walker and coworkers (12). A -BLAST search of the nonredundant data base using amino acids 279 -608 of DEFCAP identified the NBD of Nod1 and the mouse gene Mater, maternal-antigen-that embryos-require (sequence not shown). A sequence alignment of the NBDs for all of the Ced-4 family members suggests that DEFCAP is most homologous to Nod1 with 29% identity. No significant similarities between the NBD of DEFCAP and Apaf-1, DARK, and Ced-4 were found when using the NCBI  cal ribonuclease inhibitor type B (RI type B) repeats, whereas LRRs 3 and 5 share similarity to the ribonuclease inhibitor type A (RI type A) repeats. The alternating nature of the asparagine and cysteine, also known as the asparagine-cysteine ladder, in residues at position 10 of LRRs 2-6 are similar to other proteins with multiple internal LRR repeats. With the exception of leucine at position 20, the putative ␣-helical sequences of the LRRs do not share significant homology, a characteristic found in many LRRs. However, the RI type B LRRs 2, 4, and 6 share significant homology among each other as depicted in purple (Fig. 1D).

FIG. 1-continued
The CARD domain of DEFCAP is located at the most carboxyl end of the protein and is depicted in Fig. 1A by green highlighting. A -BLAST search of the nonredundant data base using the CARD of DEFCAP identifies 56% identity with ASC and 29% identity with Nod1. Less similarity is seen in an alignment comparing DEFCAP's CARD with the CARD of Apaf-1, Ced-3, and caspase-9.

Human DEFCAP mRNA Is Expressed in Multiple Tissues but Has the Highest Level of Expression in Peripheral Blood
Leukocytes and the Chronic Myelogenous Leukemia Cell Line K562-Northern blot analysis revealed DEFCAP to be expressed as at least two transcripts of ϳ7.0 and ϳ8.0 kilobase pairs in size ( Fig. 2A). Both transcripts were found in a variety of human adult tissues with the highest levels of expression in peripheral blood leukocytes, heart, thymus, and spleen. Low levels of DEFCAP mRNA expression were found in skeletal muscle, colon (no mucosa), kidney, liver, small intestine, placenta, and lung. No detectable levels of DEFCAP expression were found in the adult brain. Equal loading of RNA was determined by probing the same blots with ␤-actin cDNA. Northern analysis for DEFCAP in cancer cell lines revealed a high level of expression in the chronic myelogenous leukemia cell line K562. A weak ϳ7.0-kilobase pair band was seen in the Burkitt's lymphoma Raji, colorectal adenocarcinoma SW-480, and melanoma G-361 cell lines. No significant DEFCAP transcripts were detected in the promyelocytic leukemia HL-60, cervical carcinoma HeLa S3, lymphoblastic leukemia MOLT-4, or lung carcinoma A549.
Both DEFCAP-L and DEFCAP-S Isoforms Are Expressed in mRNAs from Normal Tissue and Cancer Cell Lines-Since the long DEFCAP isoform was cloned from a K562 cancer cell line, we first wanted to determine whether the long isoform exists in RNAs from normal tissues, and second to gain an understanding of the relative abundance of the two isoforms in various RNAs by semiquantitative RT-PCR. Oligonucleotides flanking the alternately spliced sequences of DEFCAP-L were used to amplify the long isoform as a 322-bp fragment and the short isoform as a 190-bp fragment (Fig. 3B). Both 322-and 190-bp RT-PCR products were gel-purified and confirmed to be specific to DEFCAP by DNA sequencing (data not shown). Both long and short DEFCAP isoforms were identified in RNAs from K562 cells, Jurkat cells, normal human liver, spleen, PMNs, and PBMCs. Interestingly, DEFCAP-L levels were relatively constant in all RNAs with the exception of the Jurkat and spleen RNAs that were slightly diminished. DEFCAP-S mRNA expression was weakest in K562, Jurkat, and liver but was significantly increased in spleen, PMNs, and PBMCs. ␤-Actin mRNA levels served as a control for RNA integrity and RT-PCR efficiency and were relatively constant with the exception of the K562 RNA which is slightly diminished. The weak band at ϳ590 bp seen in the K562, spleen, PMN, and PBMC lanes may represent a PCR artifact. This band was gel-purified from spleen, PMN, and PBMC RT-PCR samples, subjected to PCR with the same oligonucleotides used in the RT-PCR, and did not yield a ϳ590-bp PCR product. shown). However, an EGFP/DEFCAP-CARD fusion protein was capable of forming novel cytoplasmic filamentous structures similar to the death-effector-filaments (DEF) formed by FADD, the death-effector domain (DED-B) of procaspase-8 (14), the prodomain of caspase-2, and RAIDD (24). The formation of DEF-like structures by DEFCAP-CARD suggests that DEFCAP may have the ability to dimerize or oligomerize in a CARD-mediated manner. The fact that both full-length DEF-CAP isoforms cannot form DEFs while the CARD alone can suggests that the CARD of DEFCAP is normally in a conformation that prevents CARD-mediated oligomerization.

EGFP-DEFCAP-CARD Fusion Proteins Are Capable of Forming Death Effector Filament-like Structures in MCF7
DEFCAP-L and DEFCAP-S Bind Caspase-2 and Weakly to Caspase-To identify DEFCAP/caspase interactions, 293 cells were transiently cotransfected with either full-length DEF-CAP-L or DEFCAP-S in combination with either pcDNA3.1, caspase-2, caspase-3, caspase-8, caspase-9, or caspase-10. DEF-CAP failed to coimmunoprecipitate caspase-3, -8, and -10 and the adaptor proteins FADD and RAIDD (data not shown). However, both DEFCAP-L and DEFCAP-S were able to immunoprecipitate effectively caspase-2 (Fig. 4A). 293 cells transfected with caspase-2 alone and immunoprecipitated with Myc-AC served as a negative control and did not immunoprecipitate caspase-2 (data not shown). A ϳ48-kDa band representing the zymogen form of caspase-2 can be seen in all of the supernatant lanes but was significantly increased in samples transfected with caspase-2. The ϳ48-kDa band found in the supernatant and pellet lanes of the vector control, DEFCAP-L alone, and DEFCAP-S alone most likely represents specificity to the endogenous procaspase-2 protein. No caspase-2-specific bands appeared in the IP lanes of the vector control, DEF-CAP-L alone, or DEFCAP-S alone suggesting that we were not able to detect a DEFCAP/caspase-2 interaction with the endogenous caspase-2 protein (Fig. 4A). These results are not surprising given the fact that the endogenous caspase-2 is localized primarily to the Golgi complex (25) and the mitochondrial intermembrane space (26). Therefore, the endogenous caspase-2 protein was most likely unable to interact with overexpressed DEFCAP. However, in cells overexpressing caspase-2 and DEFCAP-L or -S, caspase-2 was effectively coimmunoprecipitated as seen by bands at ϳ48 and ϳ37 kDa. A band at ϳ33 kDa most likely representing the cleaved caspase-2 product containing the prodomain and the large catalytic subunit (27) was also found in the DEFCAP-S/caspase-2 (Fig. 4, IP lane). Interestingly, this ϳ33-kDa band was significantly diminished in the supernatant of cells cotransfected with DEFCAP-L/caspase-2 and barely detectable in the IP lane, suggesting that DEFCAP-L may be able to partially inhibit caspase-2 processing. A ϳ70-kDa protein in the IP lane denoted by an asterisk appears to be a nonspecific band that cross-reacted with the caspase-2 antibody.
To a lesser extent, DEFCAP-L and DEFCAP-S were able to coimmunoprecipitate caspase-9 (Fig. 4B). Interestingly, the major band seen in the IP lanes was the ϳ35-kDa band representing the processed caspase-9 enzyme (p35). One possible explanation for these results may be that, like Apaf-1, coimmunoprecipitation of caspase-9 by DEFCAP may lead to the autoactivation and processing of caspase-9 since the ϳ50-kDa proenzyme band is barely detectable in the IP lanes (11). Although difficult to visualize in the figure, the ϳ50-kDa proenzyme band was seen in the IP lane of cells cotransfected with both caspase-9 and DEFCAP-L but not in the IP lanes cotransfected with DEFCAP-S. Another possible explanation why we FIG. 4. DEFCAP interacts with caspase-2 and weakly with caspase-9. 293 cells were transiently transfected as described under "Materials and Methods." Supernatants (S), pellets (P), and coimmunoprecipitated proteins (IP) were analyzed by Western analysis with caspase-2-(A) or caspase-9 (B)-specific antibodies. The pellet lane was included to show that caspase-2, caspase-9, and both DEFCAP-L and -S are found in the membrane-insoluble fraction. A, arrows at ϳ48, ϳ37, and ϳ33 kDa represent the caspase-2 proenzyme and two processed forms of the enzyme, respectively. IgG heavy and light chains are depicted with arrows at 55 and 18 kDa, respectively. B, a weak DEFCAP interaction with caspase-9 is seen by a band at ϳ35 kDa representing the processed caspase-9 protein. A and B, DEFCAP expression was determined by Western analysis with Myc-horseradish peroxidase antibodies shown at the bottom of each panel.
were only able to detect the ϳ35-kDa band may be that the DEFCAP interaction is specific to the processed caspase-9 molecule. Furthermore, the weak DEFCAP/caspase-9 interaction may be indirect possibly requiring an unknown adaptor molecule. Another important observation is that DEFCAP expression levels were noticeably reduced when cotransfected with caspase-9 (Fig. 4B, bottom panels). Examination of the pellet lanes for caspase-9 alone, DEFCAP-L/caspase-9, and DEFCAP-S/caspase-9 suggests that coexpression of caspase-9 with either DEFCAP construct leads to a significant increase in the ϳ48 and ϳ35-kDa bands in the membrane-insoluble fraction. Nonspecific bands seen above the ϳ48-kDa procaspase-9 are relatively equal suggesting that protein loading does not adequately explain these differences. The functional significance of this observation remains unclear; however, DEFCAP may play a role in targeting caspase-9 to different subcellular membrane fractions.
The fact that DEFCAP interacts strongest with caspase-2 and only weakly with caspase-9 raises some interesting questions. First of all, what is the functional consequence of the ability of DEFCAP-L and -S to bind caspase-2 and can DEF-CAP interact with both long and short isoforms of caspase-2? Mice deficient of both isoforms of caspase-2 exhibit an increase in facial motor neuron apoptosis, a partial resistance of B lymphoblasts to granzyme B apoptosis, and a significant increase in the number of primordial follicles in the postnatal ovary (28). These results suggest that caspase-2 can have both a pro-and anti-apoptotic function depending on the cell type. These results raise the possibility that a DEFCAP/caspase-2 interaction can lead to either a pro-or anti-apoptotic outcome. Second, where in the cell do the DEFCAP/caspase interactions take place? Immunohistochemical and cell fractionation experiments show that caspase-2 and to a lesser degree caspase-9 have both a cytosolic and nuclear subcellular localization. Future studies are needed to investigate whether the DEFCAP/ caspase-2 or DEFCAP/caspase-9 interactions are exclusively cytosolic or whether DEFCAP is able to bind these caspases in the nucleus to exert its apoptotic function. Moreover, is DEF-CAP capable of being translocated to the nuclear membranes? This idea seems plausible given the data presented by Chen et al. (29) showing that Ced-4 translocates to the perinuclear membrane upon induction with a death stimulus.
Overexpression of DEFCAP-L, LRR/CARD-L, and LRR/ CARD-S Kills MCF7 Cells-To determine if the ectopic expression of DEFCAP constructs alone could kill cells in culture, MCF7 breast carcinoma cells were transiently transfected with NH 2 -terminal Myc full-length and mutant DEFCAP constructs (Fig. 5C). The base-line apoptosis level as determined by transfection with pcDNA3.1 was 25% (n ϭ 8, S.E. ϭ 1.655). Fulllength DEFCAP-L alone exhibited 36% (n ϭ 4, S.E. ϭ 0.958) killing, whereas full-length DEFCAP-S resulted in a 26% killing activity (n ϭ 4, S.E. ϭ 3.49), a level comparable to vector control. A lysine to serine (K340S-L) point mutation in the highly conserved P-loop of DEFCAP did not decrease the killing activity of DEFCAP-L that resulted in 38% (n ϭ 6, S.E. ϭ 2.611) apoptotic cells. These results suggest that full-length NBD of DEFCAP-L may not be functioning as an ATPase during apoptosis. Likewise, a K340S mutation in DEFCAP-S had no effect on its killing activity that remained at levels (24%, n ϭ 4, S.E. ϭ 2.743) similar to wild-type DEFCAP-S and vector control. Deleting amino acids 1-309 of DEFCAP-L (⌬PLM/PR-L) resulted in 44% (n ϭ 3, S.E. ϭ 1.649) apoptosis, a slight increase in apoptosis versus the wild-type full-length DEFCAP-L construct. A ⌬PLM/PR-S construct showed no significant killing activity, 26% (n ϭ 3, S.E. ϭ 3.936) versus vector control or full-length DEFCAP-S. A deletion construct containing only the leucine-rich repeats and the CARD (LRR/CARD-L and-S) exhibited a significant level of killing activity at 63% (n ϭ 3, S.E. ϭ 0.627) for LRR/CARD-L and 41% (n ϭ 4, S.E. ϭ 1.54) for LRR/CARD-S. These results suggest that the LRR/ CARD for both long and short isoforms may act as a constitutively active proapoptotic form of DEFCAP. In the presence of 25 M ZVAD-fmk, the pan-caspase inhibitor, the killing activity of LRR/CARD-L and LRR/CARD-S were dramatically reduced to 21 (n ϭ 4, S.E. ϭ 1.319) and 18% (n ϭ 4, S.E. ϭ 1.624), respectively (data not shown), suggesting that the mechanism of killing is most likely caspase-dependent. Base-line apoptosis with pcDNA3.1 alone in the presence of ZVAD-fmk was 11% (n ϭ 3, S.E. ϭ 0.284) (data not shown). In the presence of DEVD-fmk, the killing activity of LRR/CARD-L and -S was slightly reduced to 40 and 34%, respectively (data not shown). These results suggest that the activation of both a caspase with DEVD specificity and one with a non-DEVD specificity are required for maximal apoptotic activity by the LRR/CARD of DEFCAP.
The mutational studies also suggest that DEFCAP sequences including the NBD may act in the negative regulation of DEFCAP-L since a deletion of the PLM, PR, and NBD (LRR/CARD-L) resulted in a dramatic increase in apoptosis levels when compared with the full-length DEFCAP-L construct. Interestingly, the LRR/CARD for the short isoform (LRR/CARD-S) was also capable of inducing apoptosis but at much lower levels than LRR/CARD-L. Both LRR/CARD-L and LRR/CARD-S showed comparable levels of protein expression in transfected cells (data not shown) suggesting that these differences in killing activity between LRR/CARD-L and -S were not due to differences in protein expression. Furthermore, these results suggest that, like Apaf-1, a deletion construct containing the CARD and the domain juxtaposed next to the CARD can act as a constitutively active inducer of cell death. However, the major difference between the constitutively active Apaf-1 and that of DEFCAP is that the former is composed of a CARD/NBD and the latter is composed of an LRR/CARD. Although at present it is not clear which apoptotic signaling pathway involves DEFCAP, this study demonstrated that DEFCAP LRR/CARD constructs when transiently overexpressed were able to kill cells effectively. Future investigations into the mechanism of this killing may prove useful to understanding the role of this distinct mammalian Ced-4 molecule.