Mediation of the DCC apoptotic signal by DIP13 alpha.

DCC (deleted in colorectal cancer) is a candidate tumor suppressor gene. However the function of DCC remains elusive. Previously, we demonstrated that forced expression of DCC induces apoptosis or cell cycle arrest (Chen, Y. Q., Hsieh, J. T., Yao, F., Fang, B., Pong, R. C., Cipriano, S. C. & Krepulat, F. (1999) Oncogene 18, 2747-2754). To delineate the DCC-induced apoptotic pathway, we have identified a protein, DIP13 alpha, which interacts with DCC. The DIP13 alpha protein has a pleckstrin homology domain and a phosphotyrosine binding domain. It interacts with a region on the DCC cytoplasmic domain that is required for the induction of apoptosis. Although ectopic expression of DIP13 alpha alone causes only a slight increase in apoptosis, co-expression of DCC and DIP13 alpha results in an approximately 5-fold increase in apoptosis. Removal of the DCC-interacting domain on DIP13 alpha abolishes its ability to enhance DCC-induced apoptosis. Inhibition of endogenous DIP13 alpha expression by small interfering RNA blocks DCC-induced apoptosis. Our data suggest that DIP13 alpha is a mediator of the DCC apoptotic pathway.

The candidate tumor-suppressor gene deleted in colorectal cancer (DCC) 1 was first cloned from a locus on chromosome arm 18q where allelic deletions occur in over 70% of primary colorectal tumors (1). Since that time, loss of heterozygosity at the DCC locus and loss of DCC expression have been shown in many other tumor types (2) including prostate carcinomas (3). The loss of DCC expression has therefore been associated with tumor progression. Although the known tumor suppressor gene MADH4/DPC4/Smad4 is mapped in close proximity to the DCC locus (4,5), re-evaluation of loss of heterozygosity at 18q in colon tumor samples indicated that DCC, but not Smad4, was the most frequently altered gene on chromosome 18q13. 3-21.3 (6). In a study of 115 pancreatic and 14 biliary cancers for homozygous deletions of DCC exons and flanking 18q regions, seven homozygous deletions were seen in the region that in-cludes the DCC gene. In fact, DCC was the only known gene affected by all seven deletions. In two tumors, the deletions inactivate DCC but not Smad4 (7). These loss of heterozygosity and mutational data support the hypothesis that DCC acts as a tumor suppressor.
DCC encodes a type I membrane protein that falls into a subgroup of the immunoglobulin superfamily (8). We and others have shown that DCC may exert its tumor-suppression function through induction of apoptosis (9,10). DCC and its orthologs, UNC-40 in Caenorhabditis elegans and frazzled in Drosophila, have been established as receptors for netrin-1 and play an important role in axon outgrowth and cell migration in the developing nervous system (11)(12)(13)(14). It has been shown that induction of apoptosis by DCC can be blocked by netrin-1 (10). Expression of DCC colocalizes with areas of apoptotic precerebellar neurons in netrin-1 Ϫ/Ϫ mice, whereas apoptosis is absent in the same DCC-expressing areas in wild-type mice (15). Therefore DCC may induce cell death in settings where ligand is unavailable. Consistent with this view, netrin-1 knockout mice grow fewer cells particularly in the developing brainstem (16,17).
Very little is known about the apoptotic signaling of DCC. Unlike the other well characterized receptors such as Fas and tumor necrosis factor receptor, no apparent death domain can be identified in the DCC cytoplasmic region. Expression of bcl-2 does not abrogate DCC-induced apoptosis (9). Recently it has been reported that apoptosis induced by DCC is different from that by death receptor/caspase-8 pathway or the mitochondriadependent pathway (18). Here we report the identification of DCC-interacting protein 13␣ (DIP13␣) and present evidence showing that this DCC partner may serve as an adaptor mediating the DCC apoptotic signal.

MATERIALS AND METHODS
Cell Culture and Reagents-Colon adenocarcinoma cell line DLD1 was maintained in Dulbecco's modified Eagle's medium supplemented with 5% fetal bovine serum. All chemicals, unless otherwise specified, were obtained from Sigma-Aldrich.
Yeast two-hybrid Screen and Mapping-The LexA-based system (19) was used in our experiment. Because of a strong self-activation of the LexA-DCC cytoplasmic domain (1124 -1447) bait, we deleted the last 44 amino acids of the DCC C terminus, which contains an acidic domain. This deletion eliminated its activation activity. We used this DCC bait to screen a cDNA library constructed from HeLa cells. We also used B42-DCC cytoplasmic domain (1124 -1447) prey to screen a collection of 576 known bait proteins.
For mapping analysis, series of deletion mutants were constructed using either convenient restriction sites or PCR with appropriate primers. These mutants were assayed in the yeast system for interaction.
Additional cDNA Cloning and Sequence Analysis-A 5Ј-RACE kit (Invitrogen) was used to clone the additional 5Ј sequence. cDNA was reverse-transcribed from a total RNA sample prepared from HeLa cells using a gene-specific primer (5Ј-TGCGATTCTGTCATACGAAAGAT-GTTAT-3Ј). Another gene-specific primer (5Ј-GAACCAAGGAATCGGA-* This study was supported by National Institutes of Health Grant R01CA77489. 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) AF424738.
CAATAAATA-3Ј) was used in RACE PCR together with anchor primer obtained from the manufacturer. The 5Ј-RACE product was sequenced.
An open reading frame was identified and translated into its corresponding protein using Lasergene software (DNA* Inc.). Protein features were analyzed using the NCBI BLAST service (www.ncbi.nlm. nih.gov/) and the GCG Molecular Sequence Analysis Package (Accelrys Inc.).
Intron-exon junctions were mapped by aligning cDNA sequence with its corresponding genomic sequences in the Human Genome Project data base (www.ncbi.nlm.nih.gov/genome/guide/human/).
Construction and Expression of Fusion Proteins-All cDNA fragments were generated by PCR using high fidelity polymerases (Invitrogen) and cloned into an entry vector using the gateway cloning system (Invitrogen). Internal deletion of DIP13␣ (DIP13␣⌬428 -530) was generated by digesting DIP13 ORF with BamHI and XmaI, blunting the ends, and ligating. Expression vectors used were pDEST-TO-HisFlag (N-His-Flag tag), pDEST26 (N-His tag) and pDEST27 (N-GST tag). Transient transfection of DLD1 cells was performed using Lipo-fectAMINE Plus (Invitrogen) according to the manufacturer's instruction.
Antibodies-Anti-DCC, anti-HA, and anti-Flag antibody was purchased from PharMingen, CLONTECH and Sigma-Aldrich, respectively. Rabbit anti-DIP13␣ was made commercially using two peptides located at its C terminus and linked by 3 glycines (SSRPNQASSEggg-SQSEESDLGE).
Western Blotting and Immunoprecipitation-Two days after transfection, cells were suspended in lysis buffer (10 mM Pipes, pH 7.0, 0.5% Nonidet P-40, 80 mM KCl, 20 mM NaCl, 1 mM MgCl, 5 mM EDTA, 1 mM dithiothreitol, and protease inhibitors) and subjected to repeated freeze-thaw cycles. For Western blotting, protein extracts were resolved by SDS-PAGE, transferred onto nitrocellulose membrane, and probed with the indicated antibodies. Target proteins were visualized using the enhanced chemiluminescent method. For immunoprecipitation, lysates were incubated overnight with the indicated antibodies. Immune complexes were precipitated with protein A/G-Sepharose beads and washed with lysis buffer before being resolved on SDS-PAGE.
Transfection-Plasmid DNA was transfected with LipofectAMINE Plus (Invitrogen) according to the manufacturer's protocol.
Cell Death Analysis-Forty-eight hours after transfection, floating and attached cells were collected and enumerated by the trypan blue exclusion method. The results obtained from this method correlate well with other standard apoptosis assays used previously (9).
RNA Interference-siRNA for DIP13␣ and DIP13␤ (as a specificity control) were synthesized by Dharmacon Research Inc. The siRNA duplexes for alpha and beta, respectively, are as follows (Sequences 1 and 2).
Cells were transfected with RNA duplex and DCC expression vector using Oligofectamine (Invitrogen) following the manufacturer's protocol. Cells were recovered for Western blotting or cell death analysis 48 h post-transfection.
Microinjection-Plasmid DNA and siRNA were injected into the nucleus of HeLa cells using an Eppendorf microinjector. Injections were performed with each plasmid at a concentration of 125 ng/ml. Duplex siRNA oligos were injected at a concentration of 400 pM, resulting in an estimated delivery of 5-20 molecules/cell. An average of 200 cells was injected per condition. To quantify cell death, healthy living fluorescent cells were counted 1-2 h after microinjection, and the percent survival was determined ϳ24 h post-injection.

RESULTS
Identification of DCC-interacting Proteins-We have previously shown that the DCC cytoplasmic domain is required for the induction of apoptosis (9). To identify potential DCC signaling mediators, we used the DCC cytoplasmic domain (1124 -1447) as bait to isolate interacting proteins with the yeast two-hybrid system. Because the LexA-DCC bait protein has strong activation activity, we deleted the last 44 amino acids of the DCC C terminus, which contains an acidic domain. This deletion eliminated its activation activity. We used this DCC bait to screen a cDNA library constructed from HeLa cells. From this screening, we identified the human sina-1, sina-2, proteasome subunit p40, and six novel genes that we named DIPs (DCC-interacting proteins). In addition, the DCC cytoplasmic domain (1124 -1447) fused with the transcriptional activator B42 was used as prey to screen a collection of 576 known proteins. From this screening, we identified several proteins that interact with the DCC cytoplasmic domain in yeast, among them the Drosophila seven-in-absentia (Sina) and human FKBP12 protein. In total, 13 DCC-interacting proteins were identified. Fig. 1 shows the specific interaction of the DCC cytoplasmic domain with several selected DIPs.
Mapping of Interaction Domains-We mapped the interacting domain on the DCC cytoplasmic region for all 13 DIPs by serial deletion and yeast two-hybrid analysis. The DIP13␣interacting domain of DCC (amino acids 1240 -1273) coincided with the region (amino acids 1243-1264) that is required for DCC to induce apoptosis (10). Therefore, our subsequent studies were focused on DIP13␣. The DCC-interacting domain on the DIP13␣ was also mapped. This fragment (amino acid 454 -646) contains a phosphotyrosine binding domain (Fig. 2B).
In Vivo Interaction between DCC and DIP13-To validate the results obtained from the yeast two-hybrid system, DCC-DIP13␣ interaction was confirmed in mammalian cells by coimmunoprecipitation. Human colon adenocarcinoma DLD1 cells were transiently transfected with DCC and DIP13␣, and cell lysates were prepared 24 h post-transfection. When DIP13␣ was immunoprecipitated from the lysates by affinitypurified antibody to DIP13␣, a protein of ϳ190 kDa was coprecipitated and recognized by an antibody specific to DCC on immunoblot (Fig. 3). Reciprocally, anti-HA antibody was able to immunoprecipitate HA-DCC and co-precipitate full-length DIP13␣ (Fig. 3).
DIP13␣ cDNA and Its Genomic Structure-The DIP13␣ cDNA obtained from the yeast two-hybrid screening was considered partial because the AUG codon was not in a Kozak consensus context. A BLAST search of the human EST data base also predicted a longer gene for DIP13␣. An additional 1.6-kb 5Ј sequence was subsequently cloned via a 5Ј-RACE experiment, which identified a start codon in a good Kozak consensus context. Altogether, a 3042-base contiguous cDNA sequence was obtained, which encodes a protein of 709 amino acid residues (GenBank TM accession number AF424738). The DIP13␣ protein contains a pleckstrin homology (PH) domain (20), a phosphotyrosine binding (PTB) domain (21,22), and a coiled-coil domain (23) that partially overlaps with the PTB domain (Fig. 2B). The DIP13␣ cDNA is identical to the AKT2interacting gene, APPL (24), with the exception of a shorter 3Ј-untranslated region. Alignment of the cDNA sequence with the corresponding genomic sequence in the Human Genome Project data base predicted that the DIP13␣ gene has 22 exons spanning an ϳ36-kb genomic sequences. All intron-exon junctions conform to the GT-AG rule.
DIP13␣ Enhances Apoptosis Induced by DCC-To explore the functional relevance of DIP13␣ to the DCC apoptotic pathway, we transiently expressed the full-length DIP13␣ in DLD1 and 293T cells. We have previously shown that expression of DCC induces apoptosis in DLD1 as demonstrated by DNA fragmentation and caspase-3 activation (9). As shown in Fig.  4A, expression of DCC doubled the apoptotic population compared with vector control. This ratio of cell death was consistent with the results obtained with 293T cells (10,18

FIG. 3. Interaction of DCC and DIP13␣ in mammalian cells.
DLD1 cells were co-transfected with DIP13␣ and DCC as indicated. DIP13␣ protein was immunoprecipitated (IP) with rabbit anti-DIP13␣ antibody, and co-precipitation of DCC was detected by Western blotting (WB) with mouse anti-DCC antibody. Reciprocally, HA-DCC protein was immunoprecipitated with rabbit anti-HA antibody, and co-precipitation of Flag-DIP13␣ was confirmed by Western blotting with mouse anti-Flag antibody. cells. Co-expression of DCC and DIP13␣, however, increased the apoptotic population to almost twice as much as that caused by DCC alone and five times that of vector control (Fig.  4A). Nuclear fragmentation was seen in those dead cells (data not shown).
DCC-interacting Domain Is Required for DIP13␣ to Enhance Apoptosis-To determine whether the interaction of DIP13␣ with DCC is required for the increased cell death, we constructed a DIP13␣ deletion mutant (DIP13␣⌬428 -530) that lacks its DCC-interacting domain. This mutant lost its ability to enhance DCC-induced apoptosis (Fig. 4B). Previously, it has been shown that the DCC ligand, netrin-1, can block the ability of DCC to induce apoptosis (10). Cotransfection of netrin-1 with DCC and DIP13␣ reduced the population of apoptotic cells to a level lower than that in DIP13␣⌬-(428 -530) and DCC co-expressing cells (Fig. 4B). Interestingly, co-expression of the DCC-interacting domain of DIP13␣ (DIP13-(454 -646)) either with the full-length DCC or with the DIP13␣-interacting domain of DCC (DCC-(1240 -1273)) also induced apoptosis but to a lesser degree than that in full-length DIP13␣ and DCCexpressing cells (Fig. 4B). Expression of the DIP13␣-interacting domain of DCC (DCC-(1240 -1273)) with the full-length DIP13␣ induced a level of apoptotic cells similar to that in full-length DCC and the DIP13␣-expressing population (Fig. 4B).
Inhibition of Endogenous DIP13␣ Expression Blocks DCCinduced Apoptosis-To demonstrate that endogenous DIP13␣ can mediate DCC-induced apoptosis, an RNA interference method was used to down-regulate DIP13␣ expression. DLD1 cells were transiently transfected with siRNA for DIP13␣ or DIP13␤, another member of the adaptor protein family. DIP13␣ siRNA reduced the expression of alpha without affecting that of beta, and DIP13␤ siRNA selectively reduced the expression of beta (Fig. 5A). Similar to previous experiments, transfection of DCC doubled the apoptotic population compared with vector control. DCC together with DIP13␣ siRNA, however, reduced the apoptotic population to a level similar to that of the vector control (Fig. 5B). Because the transfection condition was optimized for small oligos but not for plasmid DNA in this experiment, we confirmed our observation by a microinjection method. Injection of DCC vector alone resulted in ϳ50% survival relative to the control vector-injected cell population. However, co-injection of DCC vector with DIP13␣ siRNA blocked DCC-induced apoptosis (Fig. 5C).

DISCUSSION
Although DCC was cloned in 1990, its function is still largely unknown. Recently, we and others have shown that DCC expression induces apoptosis and/or cell cycle arrest (9,10). To delineate the signaling mechanism of DCC-induced apoptosis, we identified several known and novel molecules that interact specifically with the DCC cytoplasmic domain. The known molecules include Drosophila sina, human Sina-1 (Siah-1), Sina-2 (Siah-2), proteasome subunit p40, and FKBP12 protein. Siah-1 has been shown to regulate DCC protein level via the ubiquitinproteasome pathway (25). It is possible that Siah-2 and proteasome subunit p40 have a similar effect on DCC, whereas the consequence of FKBP12-DCC interaction is unclear. Among the 13 DCC-interacting molecules identified, only DIP13␣ interacts with the region of DCC that is required to induce apoptosis. In the present study, we have shown that DIP13␣ interacts specifically with DCC in yeast and mammalian cells, that expression of DIP13␣ enhances DCC-induced apoptosis, that removal of the DCC-interacting domain of DIP13␣ abolishes DIP13␣'s ability to enhance DCC-induced apoptosis, and that reduction of endogenous DIP13␣ expression blocks DCCinduced apoptosis. These results indicate that DIP13␣ is a mediator of the DCC apoptotic signal. It is noteworthy that we observed the interaction between exogenous DCC (transfection with an expression vector) and exogenous DIP13␣ or exogenous DCC and endogenous DIP13␣ in DLD-1 and 293 mammalian cells by immunoprecipitation-Western blot. However, we were not able to perform an endogenous DCC and endogenous DIP13␣ interaction experiment due to the fact that most of cell lines do not express endogenous DCC. Interestingly, the neuroblastoma cell line, IMR32, expresses a high level of endogenous DCC but fails to undergo apoptosis. We were not able to detect DCC-DIP13␣ interaction by immunoprecipitation-Western blot in IMR32 cells under our experimental conditions. The lack of DCC-DIP13␣ interaction may explain why DCC does not induce apoptosis in IMR32 cells.
Recently a gene named APPL, which interacts with AKT2, has been identified (24). DIP13␣ is identical to the APPL gene. AKT is a serine/threonine kinase that can inhibit apoptosis (26). Because DIP13␣/APPL can interact with AKT, we wondered whether DCC-induced apoptosis is mediated by DIP13␣ through the interference of the AKT function. Although AKT1, -2, and -3 proteins were detected in DLD1 cells, AKT1 was more prominent than the other two (data not shown). We did not detect any changes in the total amount of AKT protein nor the phosphorylation status of threonine 308 and serine 473 of AKT in DLD1 and 293 cells expressing DIP13␣ and DCC (data not shown). Therefore, the role of AKT in DCC-induced apoptosis remains elusive.
Interestingly, Forcet et al. (18) reported that DCC induces cell death independently of either the mitochondria-dependent pathway or the death receptor/caspase-8 pathway. Moreover, DCC interacts with both caspase-3 and caspase-9 and drives the activation of caspase-3 through caspase-9 without a requirement for cytochrome c or Apaf-1. Combined with our data, this suggests that DCC, DIP13␣, and caspase-9 may coordinate DCC-induced apoptosis. It was shown that amino acids 1243-1264 on DCC were required and sufficient for DCC to induce apoptosis (10). It appeared that DCC-(1243-1264) was also required for interaction with caspase-9 (18) and for interaction with DIP13␣ (the present study). Our data further indicated that the DCC-interacting domain of DIP13␣ (DIP13-(454 -646)) was necessary for DCC to induce apoptosis. The DIP13-(454 -646) fragment can have a significant effect on apoptosis when co-expressed with either full-length DCC or DCC-(1240 -1273) (Fig. 4B). Deletion of the DCC-interacting domain on DIP13␣ (DIP13␣⌬428 -530) abolished its ability to enhance DCC-induced apoptosis. Namely, expression of DCC and DIP13␣⌬428 -530 induced the same level of apoptosis as DCC alone (Fig. 4, A and B). The apoptosis that occurs in cells expressing DCC alone was most likely caused by DCC interaction with the endogenous DIP13␣ in DLD1 cells. Indeed, reduction of endogenous DIP13␣ expression blocked DCC-induced apoptosis. Experiments are under way to determine the relationship between DIP13␣ and caspase-9 in DCC-induced apoptosis.