Deterin, a New Inhibitor of Apoptosis from Drosophila melanogaster *

Deterin, a new apoptosis inhibitor fromDrosophila melanogaster, possesses an unusual structure of only a single baculovirus inhibitor of apoptosis (IAP)-type repeat and no RING finger motif. The biochemical actions of deterin are demonstrated in SF9 and S2 cell transfection assays, in which the expressed protein acts in the cytoplasm to inhibit or deter cells from apoptosis otherwise induced by the caspase-dependent apoptosis activator reaper or by cytotoxicants. A loss of function phenotype for deterin of cell death was indicated by transfections with either a dominant negative deterin mutant or with inhibitory RNA (RNAi) for deterin. The dominant negative C-terminal fragment that antagonized antiapoptotic activity of deterin did not affect antiapoptotic activity of DIAP1 or p35. Both the baculovirus IAP-type repeat (BIR) domain and the α-helical C-terminal domain are necessary in both SF9 and S2 cells for deterin to manifest its activity to prevent cell death. The approximately 650-base deterin transcript is present in embryos, third instar larvae, and late stage nurse cells of adult females. The deterin transcript is distributed throughout early stage embryos, whereas in later stage embryos it becomes progressively restricted to the central nervous system and gonads. Whereas the nematode survivin-type IAP has thus far been implicated only as a mitotic regulator,Drosophila deterin constitutes the first invertebrate member of the survivin-type IAP group to exhibit apoptosis-inhibitory activity.

Programmed cell death is a process that specifies the elimination of superfluous or otherwise unwanted cells and tissues (1)(2)(3)(4). Observed as a programmed cell death in insect systems (5), the process has been distinguished from cell death by necrosis (6,7) and has since been conventionally characterized by cell shrinkage, blebbing of the plasma membrane, convolution of the nuclear membrane, chromatin condensation, and chromosomal DNA fragmentation into a nucleosomal ladder. In some tissues, these apoptotic cells are phagocytosed intact, without an acute inflammatory reaction. These phenomena are in contrast to cell death by necrosis, in which cells swell and lyse to release cellular contents, inducing an inflammatory reaction.
The various diverse stimuli that induce apoptosis can be external to the cell (e.g. UV, antibiotics, hormones, and liposomal cytotoxicants) or endogenous (e.g. transcriptional regulators and DNA damage due to free radicals). These apoptotic stimuli feed into apoptotic pathways of the cell by routes not well understood, although many pathways appear to converge into common or at least overlapping biochemical machinery. These pathways include activators that can activate special effector proteolytic enzymes, caspases, that in turn cleave and activate other caspases, which then cleave end target molecules; the molecular death of these target molecules causes the apoptotic death of the cell (8,9). Because many caspases in the apoptotic pathway are apparently always present in the cell, either as procaspases or perhaps performing otherwise nonapoptotic functions (10), the cell must have a means to prevent "leaky" or inadvertent triggering of the pathway. This protection is accomplished in part by the presence of inhibitors of various steps in the apoptotic machinery. For example, in vertebrates and the invertebrate Caenorhabditis elegans there are inhibitors of the "activators," such as the well known BCl-2 and related BCL-X L (11).
There are also other caspase inhibitors, such as certain inhibitors of apoptosis (IAP) 1 proteins, first discovered in baculoviruses (12) and then identified in both vertebrate and invertebrate systems (13-17). These proteins are typically characterized by BIR type repeats and a RING finger motif (18,19). However, there is much that we do not yet understand about the role and action of the IAPs generally in apoptosis and normal cell physiology. Structurally, the function of the characteristic BIR type repeat in IAPs is also little understood. Roy et al. (20) report that for the vertebrate c-IAP-1 and c-IAP-2, the BIR region alone could bind to caspases and block caspase activation. However, the inhibitory effect was greater when the RING finger was also present. More paradoxically, Hauser et al. (21) report a ubiquitin conjugating complex component that has a single BIR and no RING finger and no activity to inhibit apoptosis. Similarly, both an insect virus (AcIAP, Ref. 22) and an unrelated mammalian virus (23) encode BIR-containing proteins that have no demonstrated function to inhibit apoptosis.
Recently, a vertebrate protein, survivin, has been reported from humans and mice that represents a new subgroup of IAPs that contain only a single BIR and no RING finger (24,25). A great deal of excitement about survivin-like IAPs has been engendered by the findings that survivin is expressed in a wider variety of cancers than any other known apoptosis inhibitor (26). An apparently homologous single BIR-containing protein from C. elegans, BIR1, has also been described recently (27). The studies on vertebrate survivin demonstrated antiapoptotic activity of survivin in cell transfection assays and also suggested a possible role in cell division due to the localization of survivin to mitotic spindles and its expression in proliferating but not quiescent tissues. In contrast, genetic studies on BIR1 molecule in C. elegans indicated a role in cell division but did not detect antiapoptotic activity. These disparate findings raise the important question as to whether the antiapoptotic activity of this protein group is restricted to vertebrates, in contrast to a role solely in cell division in invertebrates (28).
To date, two Drosophila IAPs have been reported, DIAP1 and DIAP2, that contain the conventional multiple BIR and RING finger motifs (13, 16,29). Apoptosis induced by intracellular regulators reaper, grim, and hid (30 -32), and extracellular toxicants, can be inhibited by these IAPs (13, 33,34). Yet, to date, no IAP-type proteins have been published from insects that possess the single BIR/no RING finger structure. We report here the gene and cDNA for a new insect IAP, deterin, that has a single BIR and no RING finger motif. We also demonstrate that this new insect IAP exerts a distinct action to deter apoptosis, that a loss of function phenotype for deterin is cell death, and that its transcript exists in particular tissues of multiple developmental stages. As the first invertebrate member of the survivin-type IAP group to exhibit apoptosis-inhibitory activity, Drosophila deterin demonstrates that antiapoptotic activity of survivin-type IAPs is not confined to the vertebrate kingdom.

MATERIALS AND METHODS
Molecular Cloning-The cDNA clone for deterin was isolated from a Drosophila melanogaster embryo cDNA library kindly provided by Roger Brent. The sequenced cDNA was radiolabeled and used to screen D. melanogaster P1 genomic library filters (Genomesystems). The two most strongly positive hybridizing clones corresponded to independent P1 clones both associated with the 90A region of the D. melanogaster third chromosome. The inserts in these two clones were excised, recovered, and subjected to restriction digestion and probing with the radiolabeled cDNA, and the restriction pattern confirmed the overlap of the two clones. An approximately 3-kbp BamHI fragment was detected to completely contain the cDNA probe, and the deterin gene within this fragment was sequenced in both directions. During this study, a cDNA was reported into the Flybase as isolated from a late larval-prepupal library (AI260030). The reported protein coding sequence for that corresponding cDNA matched exactly the sequence of the cDNA we obtained from the embryonic library.
Mutant constructs were prepared by PCR amplification of respective coding fragments and cloning into pIE1-4 vector. Deterin deleted for its C terminus (Det-NBIR) was prepared by deleting amino acids 111-153. The construct with the C terminus alone (Det-Cterm) contained the coding region for amino acids 108 -153. In the Det-RING construct, the DIAP1 RING motif residues 391-425 were placed after residue 153 of deterin. For exchange of the C terminus of deterin with that of survivin (Det-SurvCterm), residues 111-153 of deterin were replaced with residues 98 -142 of human survivin. All constructs were cloned into the pIE1-4 insect cell expression vector.
In Situ Hybridization to Polytene Chromosomes-Salivary glands were dissected, fixed to slides, and denatured in a method similar to that described by Pardue (35). The probe (biotinylated deterin cDNA) was added to hybridization solution (formamide/SSC/dextran sulfate) and incubated with the prepared chromosomes overnight at 37°C. The slides were washed with 2ϫ SSC and then phosphate-buffered saline, and the hybridized probe was detected with Vectastain reagents (Vector).
Expression Vectors-The EcoRI/XhoHI deterin cDNA fragment was subcloned into Bluescript and then excised with BamHI/XhoHI (the latter end filled in with Klenow) and ligated into the BamHI/NruI sites of pIE1-4 vector, in which expression of the encoded protein is driven in insect cells by a baculovirus promoter ie1 (Novagen). A fusion construct was prepared by placing the coding sequence for green fluorescent protein (GFP) in frame with deterin. D. melanogaster reaper coding sequence (30) was inserted by PCR cloning into the BamHI/NruI sites of pIE1-4. The D. melanogaster apoptosis inhibitor DIAP1 reported by Hay et al. (13), which was initially named thread but which has been referred to in the literature by numerous abbreviations (DIAP1, DIAP-1, dIAP1, etc.), is referred to here by the abbreviation used in the original paper by Hay et al. (13). The coding sequence of DIAP1, reaper, the procaspases DCP1 and DRONC, and baculoviral p35 were also each cloned into the BamHI and NotI sites of pIE1-4. The control plasmid expressing ␤-galactosidase under the constitutive control of hsp70 promoter (phsp-␤-gal) was as described (38).
Cell Transfections-Spodoptera frugiperda SF9 cells were maintained at 25°C in SF900 SFM media (Life Technologies, Inc.). D. melanogaster S2 cells were maintained at 25°C in Schneider's Drosophila medium, (Life Technologies, Inc.) with 10% fetal calf serum. The pIE1-4-deterin construct (above), and the pIE1-4 vector only, were each separately and stably transformed into SF9 cells by cotransfection with pIE1-neo plasmid, and selection with neomycin, as per manufacturers instructions (Novagen). For transfections involving Drosophila reapermediated apoptosis, 10 6 cells were transferred to 6 well culture dishes (Corning) in 1 ml media, and transfected with 3 l of Cellfectin (Life Technologies, Inc.), with 0.2 g of reporter phsp-␤-gal or the pIE1-4-GFP construct, as noted, and with either 2 g of pIE1-4-reaper and/or 6 g of pIE1-4-deterin, pIE1-4-GFP-deterin, pIE1-4-DIAP1, or pIE1-4-p35. In addition, pIE1-4 vector was added in an amount sufficient to equalize the total amount of DNA transfected in each treatment to 8.2 g. To assess deterin activity against exogenous apoptotic activation, liposomal conditions for induction of caspase-dependent apoptosis were used (39 -41). We determined that these conditions of S2 and SF9 cells were an excess of liposomal reagent relative to numbers of treated cells of 3 l of Cellfectin to 0.5 ϫ 10 6 cells/ml. For RNAi experiments, sense and antisense transcripts were synthesized from the corresponding cDNA in Bluescript SKϩ vector, using T3 or T7 RNA polymerase. The resulting transcript preparations were mixed, briefly boiled, and then annealed. The given RNAi was subjected to the same transfection procedures as transfected DNAs.
For all transfections, 5 h after addition of reagents, the appropriate medium was added to achieve a total volume of 2 ml in each well, and 24 -48 h later (as noted), the cells were either stained for ␤-galactosidase activity (2 mM MgCl 2 , 5 mMK 4 Fe(CN) 6 -3 H 2 O, 5 mM K3Fe(CN) 6 , 0.2% 5-bromo-4-chloro-3-indoyl ␤-D-galactoside) or fluorescence-activated cell sorter analyzed for the GFP reporter, as described above. For the ␤-galactosidase reporter, the number of positive reporter (blue X-gal staining) cells in the consecutive fields of view in a transect across the widest diameter of each well were counted (typically several hundred to several thousand). The standard method for assessing cell viability used previously for survivin and other IAP-type molecules was also used here (22,24,32,42,44,45) (i.e. measurements are made within a time course such that the number of surviving cells (those not excluding trypan blue or those expressing the reporter) have had insufficient time to significantly proliferate); this procedure thereby prevents any bias towards showing a higher than actual number of surviving viable cells. This procedure was additionally verified in the present studies by preliminary tests, which showed little cell division by transfected SF9 or S2 cells during the time course under the conditions used and which showed that there was no increase in DNA synthesis during the time course in cells overexpressing deterin relative to control cells. In all transfection experiments, each treatment was repeated at least three times. Data are reported as mean and standard error.
Northern and RT-PCR Transcript Analysis-RNA was extracted from 0 -4-h-old embryos, 4 -8-h-old embryos, and late feeding stage final instar larvae, as described (46). Embryos or larvae were homogenized in SDS lysis buffer and digested with proteinase K, and the nucleic acids were extracted with phenol and then with phenol/chloroform and ethanol-precipitated. After fractionation on formaldehydecontaining agarose gels, the nucleic acids were transferred to Nytran membrane (Schleicher and Schuell) and probed with [␣-32 P]dCTP-labeled deterin cDNA; radiolabeling was performed by the random primer method. Signals were visualized by autoradiography. RNA samples from embryos or larvae were also used as the RT-PCR template, using primers corresponding to the 5Ј end and 3Ј end of the protein-coding region, respectively. Templates for RT-PCR were RNAs extracted from 0 -12 h old embryos, late feeding stage third instar larvae, and mated females. For the transcript corresponding to the cloned deterin cDNA, the expected PCR product using these primers is 459 base pairs in length. In order to confirm the presence of endogenous Drosophila deterin transcripts in Drosophila S2 cells, cultured cells were collected and boiled for 5 min, the debris was pelleted, and aliquots of the supernatant were used as template in various PCRs using combinations of primers designed from cDNA designed to yield diagnostic and different size transcripts in each reaction. Confirmation of the presence of Drosophila deterin transcripts in Spodoptera SF9 cells stably transformed with a Drosophila deterin-expressing construct was performed on the cultured cells in a similar manner. The presence of overexpressed deterin protein in these SF9 cells was also confirmed by immunoblotting cellular proteins after SDS-polyacrylamide gel electrophoresis using rabbit polyclonal antibodies prepared against bacterial recombinant Drosophila deterin.
In Situ Hybridization-Adult female ovaries were fixed with 4% paraformaldehyde in phosphate-buffered saline, pH 7.4, permeabilized with 4 g/ml proteinase K, refixed, prehybridized, and then hybridized in hybridization solution: 50% deionized formamide, 5ϫ SSC, 100 mg/ml each of sonicated salmon sperm DNA and yeast type X tRNA (Sigma), 50 mg/ml heparin, and 1%Tween. Hybridization was performed overnight at 65°C, with either sense or antisense riboprobe of the deterin cDNA, labeled by incorporation of digoxigenin-labeled UTP. After washings at 65°C, alkaline-phosphatase conjugated antidigoxigenin antibody (BMB; 2 l per 4 ml of phosphate-buffered saline/0.1% Tween) was added for 1 h of incubation at room temperature. The preparation was washed with pH 9.0 Tris-buffered saline/0.1% Tween and then stained with alkaline phosphatase in the same buffer.
Mixed stage embryos were dechorionated with bleach; agitated between n-heptane and 4% formaldehyde, 0.1 M Na 2 PO 4 ; passed sequentially through methanol, ethanol, xylene/ethanol, and methanol; and finally fixed in 5% formaldehyde in phosphate-buffered saline, pH 7.4, 0.1% Tween (PBT)). After the embryos were changed to PBT, they were digested with 40 g/ml proteinase K (BMB) for 8 -10 min and then postfixed in 5% formaldehyde. The formaldehyde was washed out with PBT, and the embryos were prehybridized for 1-2 h at 55°C in the same hybridization solution as described above. Hybridization in the same buffer overnight at 55°C was performed with the same sense or antisense probes as used for the ovary in situ hybridizations. The next day, the embryos were washed with PBT, incubated with alkalinephosphatase-conjugated antidigoxigenin antibody, and stained by methods similar to that described above for the ovaries.

Functional Structure of the Deterin Gene and Its Encoded
Protein-The encoded sequence for deterin, a 153-amino acid protein, contains a BIR-type repeat that is typical of BIRs found in IAP-type apoptosis inhibitory proteins (Fig. 1). However, it possesses a single BIR and no RING finger motif, whereas most IAP-type proteins encode two or three BIR re- The solid and open arrows indicate the positions of the two introns that disrupt the protein coding sequence of the corresponding region of the gene. Note that the second intron shown is inserted at exactly the predicted junction between commencement of a strongly ␣ helical region and the preceding primarily coiled/␤-form region. D, viability of SF9 cells was decreased by transfection with a plasmid expressing the established apoptosis inducer, Drosophila reaper (rpr). Cell viability was rescued by cotransfection with either a plasmid expressing deterin (rpr ϩ Det) or a plasmid expressing the established apoptosis inhibitor, Drosophila DIAP1 (rpr ϩ DIAP1). Percentage of nonviability was calculated in reference to the number of viable cells expressing the ␤-galactosidase reporter in the controls that were transfected with an equivalent amount of empty expression vector only. peats, and most have a RING finger. In this structure, the encoded protein is most similar to a human and mouse apoptosis inhibitor, survivin and TIAP, respectively (25,42), although there is considerable divergence in structure in the N-terminal 25 amino acids, including a 13-residue insertion, and little sequence homology over the C terminus (Fig. 1B). Cotransfection of SF9 cells with a deterin-expressing plasmid rescued the cell viability from the apoptotic effects of reaper, the insect cell death activator (Fig. 1D).
The gene encoding deterin was isolated in order to assess indications from gene structure of the functional organization of the protein. The deterin cDNA was used for polytene chromosome in situ hybridization, and a single specific signal was detected at band 90A1-A2, on the right arm of the third chromosome (not shown). The cDNA was also used to probe a P1 filter genomic library. Both of two overlapping, approximately 80-kilobase pair specific hybridizing clones were shown by restriction mapping to encode the same single deterin sequence consistent with the in situ results. On this basis, it does not appear that sequences closely similar to the single detected deterin gene are located at loci outside of 90A1-A2. The transcription start site has not been mapped, but the protein coding sequence of the cDNA is the same as an expressed sequence tag sequence deposited into the BDGP Flybase data base (accession number AI260030). The protein coding sequence is interrupted by two introns of 77 and 61 base pairs and is flanked in the mature transcript by a 5Ј untranslated region and a 3Ј untranslated region of 91 bases (Fig. 1A).
The locations of the introns may also be informative in relation to the functional structure of the deterin protein. Despite the divergence in primary structure of the human survivin and the D. melanogaster deterin, the first intron in both genes is inserted in essentially identical positions (Fig. 1B, closed arrow). Both first introns begin either immediately before (deterin) or immediately after (survivin) the codon for the basic amino acid at the beginning of the most conserved motif between the two protein sequences (2 50 KMAEAGFYW 58 for deterin, 37 R2MAEAGFIH 45 for survivin). In addition, predictive secondary structural analysis for the encoded deterin protein suggests that the first exon encodes a protein region that is predominated by ␣-helical conformation (Fig. 1C). In contrast, the second exon is predicted to encode a region nearly devoid of ␣-helical structure and instead has a primarily coiled coil conformation, along with most of the ␤-form present in the protein. Furthermore, the second intron is positioned at exactly the residue (Val 111 ) predicted to mark both the end of the BIR repeat and the beginning of a final domain with a high helical density (Fig. 1, B and C). This final domain was confirmed (as described below) as possessing a discrete biological activity relating to cell death.
Expression of Deterin Transcript-Nurse cells undergo apoptosis only after they have completed their function of delivering molecules and cytoplasmic components to the developing oocyte (48). The specific developmental timing of this apoptosis suggests a mechanism to prevent premature apoptosis in oocytes prior to that time. We analyzed the abundance of the deterin transcript in the ovaries of the adult female by in situ hybridization. The level of deterin mRNA increased sharply in later stage nurse cells, especially during stages 10 and 11 (Fig. 2D). The transcript accumulates in the darkly staining cytoplasm of the nurse cells, shortly before the dumping of cytoplasmic contents into the developing oocyte.

FIG. 2. Expression of deterin transcript.
A-C, in situ hybridization to deterin transcript in embryos. The cDNA for deterin was used to probe various stage embryos. Subsequent to the initial wide distribution of the signal in prezygotic transcription stages 1-4 (not shown), the positive signal became more restricted as development proceeds from stage 5 (A) through stage 10 (B; arrow points to brain) and to stages 15 and 16 (C; arrows point to gonads). D, in situ hybridization to deterin transcript in adult female ovaries. Stage 10 and 11 follicles (arrows) show a much stronger cytoplasmic signal than do earlier stage follicles. Control (sense strand) probe did not show this specific staining. E, Northern analysis to determine size of deterin transcript. The resultant signals show that in each stage, the detected transcript was 600 -700 bases, consistent with the size of the deterin cDNA that we isolated from an embryo cDNA library. Amounts of RNA loaded were affected by the recovery of RNA obtained from each preparation, and being used only for transcript size determination are thus not inferred to reflect stage differences in abundance. F, RT-PCR analysis of the presence of the deterin transcript in embryonic (E) and late third instar feeding stage (L) and mated adult female (A) stages. The primers used for the 5Ј and 3Ј ends of the deterin protein coding sequence would be predicted to yield a 459-base pair product, as was detected in each of the reactions.
The presence of abundant deterin transcript in the nurse cells just prior to dumping of their cytoplasmic contents into the oocyte suggested that the transcript may be found in the embryo. In situ hybridization detected the deterin transcript distributed throughout the embryo prior to stage 4 (not shown), consistent with a putative maternal source. As embryonic development progresses after the onset of zygotic transcription ( Fig. 2A), the transcript shows more restricted patterns, with particular abundance in the brain and nervous system (Fig. 2B). By stage 16, the transcript is also concentrated in the gonad (Fig. 2C), this being a highly mitotically active tissue during D. melanogaster development (e.g. 49). (Negative sense strand control did not show these patterns of staining (not shown).) A single size of the transcript, approximately 650 bases, was observed in RNA from both 0 -4-h and 4 -8-h embryos (Fig.  2E). This size is consistent with our retrieval from the cDNA library of a deterin clone of 600ϩ bases, not including the poly(A) tract. A similarly sized transcript was also observed in RNA from late feeding stage larvae. An independent assessment of the presence of the deterin transcript was made with RT-PCR using RNA extracted from 0 -12 h embryos, late feeding stage third instar larvae, and mated adult females. As shown in Fig. 2F, the expected 459-base pair product was detected for all three stages, and no larger product was detected. This result suggests that these stages do not produce transcripts related to deterin that contain additional BIRs or RING finger coding sequence between the extreme 5Ј and 3Ј protein coding sequences used to fashion the PCR primers. Similar PCR analysis using several different pairs of Drosophila deterin primers demonstrated the presence of deterin transcripts in the Drosophila S2 cells (Fig. 3C) used in functional assays, as described below.
Functional Activity to Deter Cell Death-The activity of deterin to prevent cell death (Fig. 1D) was further assessed in both dipteran (Drosophila S 2 and K c ) and lepidopteran (Spodoptera SF9) cells, using two independent end point measures (fluorescence-activated cell sorter sort for GFP, or a ␤-gal reporter), in both deterin-and GFP-deterin overexpression systems. In addition, two different methods of inducing cell death were employed (the natural apoptosis activator reaper and a liposomal cytotoxicant). Reaper has been well established to induce cell death by promotion of the apoptotic cell death program (31), and cells toxified by liposomal agents also undergo apoptosis, as indicated by the established markers of cell shrinkage and fragmentation of chromatin (39 -41). Cotransfection of a pIE1-4 vector expressing Drosophila GFP-deterin into either reaper-transfected S 2 or K c cells resulted in significantly greater survival of ␤-gal reporting cells as compared with only cotransfection of the empty vector (Fig. 3A). In addition, a GFP-deterin fusion-expressing construct effectuated greater survival of ␤-gal reporting cells against liposome (Cellfectin)-induced cell death in SF9 cells, verifying that the GFP-deterin fusion possesses apoptosis-inhibitory activity as does deterin (Fig. 3B). The GFP-deterin fusion was localized primarily in the cytoplasm (Fig. 3D), and thus its activity in these assays is apparently exerted through cytoplasmic biochemical machinery.
The actual decrease in number of viable cells under liposome-induced cell death conditions was confirmed with Drosophila S2 cells. From 24 to 48 h after exposure to liposome cytotoxication without deterin transfection, the number of viable S2 cells decreased by 50%, whereas there was no decrease in viable cell number in exposed cells that were also transfected The activity of deterin to deter cell death relative to other specific inhibitors of caspase-dependent apoptotic pathways was also investigated. The baculoviral protein p35 specifically inhibits apoptotic cell death by way of inhibition of the caspases that are activated during cytotoxin-induced apoptosis (17,51), and DIAP1 specifically inhibits apoptosis by way of interaction with apoptosis-associated caspases (52). In both SF9 and S2 cells exposed to liposomal conditions of apoptosis, cells transfected with the plasmids expressing the caspase-inhibiting p35 or DIAP1exhibited significantly greater survival than those transfected with pIE1-4 vector only (Fig. 4, A and B). In parallel transfections, cells transfected with deterin-expressing plasmids exhibited a survival similar to or greater than that shown by those transfected with the caspase-inhibiting p35and DIAP1-expressing plasmids (Fig. 4, A and B).
The antiapoptotic action of deterin was also confirmed using a cell line stably transformed for deterin expression. Those cells were confirmed by RT-PCR and immunoblotting to express the deterin transcript and protein (Fig. 3C) and were significantly more resistant to Cellfectin-induced cell death than was the control cell line stably transfected with vector only (Fig. 4C).
Functional Structure of Deterin-The secondary structure predictions and the placement of the second intron (Fig. 1, B and C) suggested that deterin may possess a discrete functional C-terminal domain. Indeed, deterin truncated so as to delete the C-terminal showed distinctly reduced antiapoptotic activity (Fig. 4D). Additional mutational analysis showed that deletion of the N-terminal region, 5Ј to the BIR domain, had less effect on activity. Addition of a prosthetic RING finger (from DIAP1) or substitution of the C-terminal domain for the homologous domain of human survivin again distinctly reduced antiapoptotic activity (Fig. 4D).
The functional participation of the C-terminal domain in affecting apoptotic pathways was further examined in S2 cells using established Drosophila and baculoviral inhibitors of apoptosis. Under liposomal conditions that induce apoptosis, cells transfected with plasmids expressing full-length deterin exhibited a significantly greater survival that those transfected with empty vector alone (Fig. 5A). However, when a plasmid expressing only the C terminus of deterin was transfected, cell survival was reduced to even lower than that for transfection of only the empty plasmid vector (Fig. 5A), suggestive of a dominant negative activity against endogenous deterin. This hypothesis was supported by the result that when the C terminus expressing plasmid was cotransfected along with the plasmid expressing the full-length deterin, the ability of the full-length transfected deterin to prevent cell death was strongly reduced. That this dominant negative effect was specific to deterin action and not just an artifact of protein overexpression was demonstrated by cotransfections of the C terminus expressing plasmid with a plasmid expressing either of the caspase-inhibitors DIAP1 or p35. The ability of DIAP1 or p35 to prevent liposome-induced cell death was unaffected by cotransfection with the plasmid expressing the deterin C terminus (Fig. 5A).
These results indicate a model of deterin function in which the biochemical site(s) of deterin action is not coextensive with those of DIAP1 or p35.
In addition to the protective effect offered by overexpression of exogenous deterin, the apparent dominant negative activity of the C-terminal fragment of deterin in S2 cells indicates that natural, endogenous full-length deterin exerts a normal activity that prevents cell death. (Untransfected S2 cells were ver-

FIG. 4. Wild type deterin antiapoptotic activity in comparison to DIAP1 (A) and to p35 (B).
In C is shown the greater survival of cells stably transformed with pIE1-4 trx-His-s-deterin fusion protein as compared with those transformed with only the pIE1-4 vector. In D, the uppermost indicated construct is the wild type. Beneath it are shown a mutant deleted for the C terminus (Det-NBIR), a mutant deleted for all but the C terminus (Det-Cterm), a mutant in which the DIAP1 RING motif is placed onto the deterin C terminus (Det-RING), and a mutant in which the deterin C terminus domain is exchanged for the counterpart domain from human survivin (Det-SurvCterm). In all four panels, the cells were subjected to liposomal cytotoxication that could be inhibited, as shown, by the established caspase inhibitors p35 and DIAP1, as well as by deterin.
ified by PCR analysis to express the endogenous deterin gene, Fig. 3C.) We therefore conducted additional tests on the cellular phenotype of a deterin loss of function, using a variation of the RNAi loss of function method used previously for Drosophila (53). We first verified the specificity of action of an RNAi on its cognate target gene in transfected cells. SF9 cells that express endogenous deterin (Fig. 3C) were transfected with a GFP-expressing reporter plasmid, along with either double stranded RNA encoding ␤-galactosidase (RNAi ␤-gal) or the same amount of control pIE1-4 vector DNA. After 48 h, counting of GFP-positive cells by fluorescence-activated cell sorter verified that the percentage of cells expressing the GFP reporter was essentially unaffected by the transfected RNAi-␤gal (Fig. 5D). In a parallel series, cells transfected with a different target plasmid expressing ␤-galactosidase, along with either RNAi-␤-gal or control pIE1-4 vector, were stained for activity of the target ␤-galactosidase. As shown in Fig. 5E, expression of the target ␤-gal was strongly inhibited by the cognate RNAi-␤-gal, as compared with ␤-gal expression in cells that were instead transfected with the control pIE1-4 DNA. The data from these two experiments confirm the high specificity of the inhibitory action of the RNAi on expression of its cognate target. When this specific loss-of-function technique was then applied to deterin, we observed that transfection of cells with RNAi-deterin, to inhibit endogenous deterin expression, strongly reduced cell survival. When the transfected RNAi-deterin was titrated by cotransfection with a deterinexpressing plasmid, cell survival was significantly increased (Fig. 5F). In an independent approach, SF9 cells stably transformed with either a deterin-expressing construct or with only the empty pIE-14 vector were each transfected with RNAideterin. The cell line transformed with the deterin-expressing construct exhibited a 5-fold resistance to the RNAi-deterin effect, as compared with the cells stably transformed with the vector only (not shown). Both the C terminus dominant nega-tive tests and the RNAi results indicate a loss of function phenotype for deterin of increased cell death.
Because many apoptotic activators stimulate caspasedependent apoptotic pathways, we examined the relative effect of deterin versus p35 to inhibit cell death induced by two Drosophila caspases, DCP-1 and DRONC (54,55). As shown in Fig. 5C, cotransfection of a p35-producing plasmid was significantly more effective than a deterin-expressing plasmid at inhibiting cell death caused by transfection with plasmids encoding either procaspase DCP-1 or procaspase DRONC.

Functional Structure of Deterin Gene and Encoded Protein-
Until now, only two IAP-type apoptosis inhibitors have been functionally identified from Drosophila, DIAP1 and DIAP2 (Refs. 13, 16, and 48; for reviews, see Refs. 18, 28, 52, and 56). Both of these previously reported inhibitors possess the classical IAP signature motifs of two or more BIR-type repeats and a C-terminal RING finger (57). The structure of the encoded deterin is in marked departure from DIAP1 and DIAP2 in that it is missing the RING finger and has only a single BIR motif. This unusual arrangement is similar to that of vertebrate survivin recently described from humans and mouse (25,42). Although some other vertebrate cellular proteins containing a single BIR motif have been reported (e.g. 21, 23), they have not been shown to affect apoptotic pathways, and no such other BIR-containing proteins have been reported from Drosophila (1,18). Our screening of a P1 Drosophila genomic library did not show any other specifically hybridizing loci outside the 90A1-A2 interval, and hybridizations detected no other sequences closely related to deterin within two overlapping P1 clones of approximately 80 kilobase pairs containing the deterin coding sequence. In addition, only a single deterin-like expressed sequence tag sequence has been reported into the BDGP Flybase (AI260030). So far, the Drosophila genome pro- ject has not reported a deterin sequence from other loci, and there is no reported p-element insertion in this locus. Thus, at present, there is no evidence that other deterin-like sequences exist in the Drosophila genome.
The deterin gene contains two introns that disrupt the protein coding sequence in an arrangement suggestive of a functional organization for the protein. The first intron of the D. melanogaster deterin gene is located to within one codon of the corresponding position of the first intron in the human survivin sequence, i.e. straddling the conserved basic residue that precedes the highly conserved motif MAEAGF. The first intron of the survivin-like C. elegans gene has been ascribed to the position immediately 3Ј to the codon for the conserved G in the above motif (59). The second intron in the deterin gene is inserted exactly between the end of the BIR and the beginning of the C-terminal domain that is predicted to initiate a region of predominantly ␣ helical structure. There have been several analyses on the functional structure of BIR domains for those IAP-type proteins that contain several BIRs and a RING finger. Mutant human XIAP deleted for the RING motif and all but a single BIR is still capable of inhibiting apoptosis (60). In insect systems, some baculoviral IAPs require both BIR domains and a RING finger for antiapoptotic activity (22,34). However, a recent analysis showed that BIR2 from either a baculoviral (Op-IAP) or cellular (DIAP1) IAP is sufficient and necessary to block apoptosis induced by the insect apoptotic regulator HID (33).
Deterin Structure Necessary for Inhibition of Programmed Cell Death-At present, little has been reported about the functional structure of IAP-type proteins that possess a single BIR, such as vertebrate survivin, mouse TIAP, or nematode BIR-1. Human survivin deleted for the ␣-helical C terminus does not bind to mitotic spindles, where it is postulated to have a function in regulation of cell division and to which it must bind in order to exert its antiapoptotic activity (24). Also, transgenic expression of human survivin in nematodes could only partially rescue failed cytokinesis in nematodes not expressing their BIR-1 gene (27). However, the necessity for the BIR1 domain or the ␣-helical C terminus for antiapoptotic activity of survivin-type molecules is not fully clear, especially in view of the existence, until now, of a question as to whether the invertebrate survivin homologs even have antiapoptotic activity.
The data from the present study demonstrate the activity of deterin to inhibit apoptosis in insect cells. Neither the isolated ␣-helical C terminus alone nor deterin deleted for the C terminus alone is able to exhibit the anti-apoptotic activity of the full-length protein. However, the C terminus appears to be a functional domain, because expression of the C-terminal fragment interferes with the cell survival promoting-action of fulllength deterin. Addition of the DIAP1 RING to the C terminus of deterin distinctly suppressed the deterin antiapoptotic activity, suggesting that the RING domain may negatively regulate deterin antiapoptotic activity. This result in a cell transfection system compliments the earlier report from transgenic flies that removal of the RING domain from DIAP1 increases its antiapoptotic activity (16). We also observed in this study that substitution of the human survivin C terminus for the deterin C terminus markedly reduced the antiapoptotic activity of deterin. Thus, despite the shared apparent ␣-helical structure of the two domains, including several conserved residues, there are significant differences between them that prevent the human sequence from functioning in antiapoptotic action in insect cells. Also, full-length human survivin was not able to fully substitute for the apparent C. elegans homolog, BIR-1, in cytokinetic functions of BIR-1 in nematode cells (27). Thus, although two actions of survivin in humans (mitotic regulation and inhibition of apoptosis) are expressly exhibited by either the invertebrate BIR-1 protein (mitotic regulation) or invertebrate deterin (inhibition of apoptosis), the necessary structural features of the respective protein in humans are not synonymous with the necessary structural features for the same activities in invertebrates. Cellular Death Activities Interceded by Deterin-Recently, it was shown that human survivin is capable of inhibiting apoptosis induced by anticancer agents or chemicals such as Taxol, apparently in part by binding to or preventing the proteolytic activation of downstream caspases 3 and 7 but not of upstream caspase 8 (24,61,62). Such a downstream action would predict that survivin-type regulators could also inhibit apoptotic signals initiated by endogenous as well as exogenous activators that feed into the caspase-dependent pathway. Supporting this hypothesis is the demonstration here of the ability of deterin overexpression to block apoptosis that is induced by either the natural caspase-dependent regulator reaper or by a liposomal cytotoxicant, the cell death effect of which is turn inhibitable by two regulators known to inhibit caspase-dependent apoptosis. Reciprocally, very important in fashioning hypotheses about deterin function in living cells is identification of a loss-offunction phenotype. The results of the two very different lossof-function tests, the dominant negative C-terminal fragment of deterin and RNAi-deterin, caused decreased cell survival, indicating a necessary role for deterin in cell survival.
Although deterin, DIAP1, and p35 all possess activity to inhibit apoptotic pathways induced by either reaper or a cytotoxic liposome, the biochemical site(s) of deterin action is not coextensive with that of either DIAP1 or p35. The deterin C terminus is capable of interfering with the antiapoptotic action of cotransfected full-length deterin, but it had no detectable effect on cotransfected DIAP1 or p35. Neither the cellular apoptosis inhibitors DIAP1 and DIAP2 nor the viral p35 contains the C-terminal domain as possessed by deterin, further consistent with a model for biochemical targets of deterin that are not coextensive with these established apoptosis inhibitors. In addition, cotransfection of DIAP1 along with deterin did not increase cell survival over that effected by transfection of deterin alone. Also, transfection of a p35-expressing plasmid was much more effective at inhibiting apoptosis induced by cotransfection of procaspase form of either DCP1 or DRONC. These parameters of deterin action suggest a new pathway for regulation of insect apoptotic mechanisms that converge into the caspasedependent cell death program, a new pathway that necessarily involves a discrete functional activity of the deterin C terminus domain in biochemical targets not reached by either cellular DIAP1 or DIAP2 or viral p35.
Cellular and Developmental Expression of Deterin-We have shown that deterin is expressed in proliferating cell culture of Drosophila S2 cells and that it is expressed in tissues that proliferate during development (e.g. gonad). The related vertebrate survivin is expressed in all proliferating cell lines examined (61), is expressed in a cell cycle-dependent manner (24), is expressed in proliferating tissues (57,63), and is bound to mitotic spindles in a manner that is disrupted by either deletion of or dominant negative overexpression of its C-terminal domain (24). Those data generated the proposition that survivin in some manner connects apoptosis regulation with the regulation of cell division and proliferation (64). Recently, a yeast BIR-containing protein was demonstrated to be necessary for proper cell division, suggestive of an ancestral role for BIR-containing proteins in mitosis, with antiapoptotic activities being evolutionarily derived (50,58,59).
Although the antiapoptotic activities of Drosophila DIAP1 and DIAP2 have been assessed by a number of laboratories, little is known about the normal developmental expression of IAPs in insects. The cDNA for DIAP1 has been recovered from a third instar cDNA library, but the temporal and tissue distribution of the transcript within embryos is not reported (13). Very recent reports have detected DIAP2 but not DIAP1 transcripts in late larval salivary glands (47) and both DIAP1 and DIAP2 in the female ovary (48). The deterin transcript is present throughout the embryonic development of Drosophila, initially widespread and later becoming progressively restricted in its distribution. During late embryonic stages, it is most abundant in the gonad and certain brain cells. Recently, proteins regulating cell cycle have shown a similar embryonic expression pattern. For example, small-minded (smid) is initially expressed throughout the embryo, but after the extended germ band stage, it is restricted to neurogenic ectoderm and gonad. smid null mutants exhibit an abnormally small central nervous system, due to defective mitosis of postembryonic neuroblasts and their subsequent apoptotic death (49). Our Northern and RT-PCR data indicate that the only, or the principal, deterin-related transcript occurring during the embryonic and late larval stages is the deterin transcript cloned here.
In the adult female, we found deterin to accumulate in high abundance in the cytoplasm of late stage nurse cells, with a maximal level just before the nurse cells dump their cytoplasms into the oocyte. The high abundance of the transcript just prior to dumping and the apparent presence of the transcript in the embryo prior to the start of zygotic transcription is indicative of a maternal contribution of deterin to the embryo. The high level in the nurse cells may also reflect an antiapoptotic role of deterin in the nurse cells themselves. Although DIAP1 and DIAP2 have been shown to inhibit apoptosis caused by expression of reaper, grim, and hid in transgenic compound eyes (16) or in transfected cells (33), there is current uncertainty as to what other apoptosis regulator(s) are required for the precise control of nurse cell death, and the natural antiapoptotic mechanism(s) holding that apoptosis in check until that stage is not known (43,48). Other apoptotic activities have been reported that are not promoted by reaper, grim, or hid, and the inhibitory regulators modulating those activities are also unknown (31). Our detection of deterin in adult female nurse cells, embryos, and last instar larvae suggests that is likely to have importance as a regulator in a number of different tissues throughout insect development. Although transpositions, mutations, or deficiencies yielding antiapoptotic or cell proliferative phenotypes have not yet been reported into the Flybase for the 90A1-A2 interval, the power of the Drosophila genetic system may offer opportunities to rapidly discern basic mechanistic actions of deterin that are also of important clinical significance.