Host Insect Inhibitor-of-Apoptosis SfIAP Functionally Replaces Baculovirus IAP but Is Differentially Regulated by Its N-Terminal Leader

ABSTRACT The inhibitor-of-apoptosis (IAP) proteins encoded by baculoviruses bear a striking resemblance to the cellular IAP homologs of their invertebrate hosts. By virtue of the acquired selective advantage of blocking virus-induced apoptosis, baculoviruses may have captured cellular IAP genes that subsequently evolved for virus-specific objectives. To compare viral and host IAPs, we defined antiapoptotic properties of SfIAP, the principal cellular IAP of the lepidopteran host Spodoptera frugiperda. We report here that SfIAP prevented virus-induced apoptosis as well as viral Op-IAP3 (which is encoded by the Orgyia pseudotsugata nucleopolyhedrovirus) when overexpressed from the baculovirus genome. Like Op-IAP3, SfIAP blocked apoptosis at a step prior to caspase activation. Both of the baculovirus IAP repeats (BIRs) were required for SfIAP function. Moreover, deletion of the C-terminal RING motif generated a loss-of-function SfIAP that interacted and dominantly interfered with wild-type SfIAP. Like Op-IAP3, wild-type SfIAP formed intracellular homodimers, suggesting that oligomerization is a functional requirement for both cellular and viral IAPs. SfIAP possesses a ∼100-residue N-terminal leader domain, which is absent among all viral IAPs. Remarkably, deletion of the leader yielded a fully functional SfIAP with dramatically increased protein stability. Thus, the SfIAP leader contains an instability motif that may confer regulatory options for cellular IAPs that baculovirus IAPs have evolved to bypass for maximal stability and antiapoptotic potency. Our findings that SfIAP and viral IAPs have common motifs, share multiple biochemical properties including oligomerization, and act at the same step to block apoptosis support the hypothesis that baculoviral IAPs were derived by acquisition of host insect IAPs.

Apoptosis is a prevalent host cell response to virus infection. Representing an important antivirus defense, apoptotic cell death can limit multiplication and virus dissemination in the host. Thus, the mechanisms by which a host organism detects a viral intruder and initiates the apoptotic response are critical to the outcome of the infection for both the host and virus. The cellular inhibitor-of-apoptosis (IAP) proteins are important candidates for sensing virus infection and determining cell fate by virtue of their central position in the apoptosis pathway (reviewed in references 35, 36, and 44). Affirming their importance in regulation of apoptosis, IAPs are encoded by multiple DNA viruses, including baculoviruses, entomopoxviruses, iridoviruses, and African swine fever virus (reviewed in 3). Nonetheless, the molecular mechanisms by which viral IAPs regulate virus-induced apoptosis and how they biochemically differ from cellular IAPs are poorly understood.
The IAPs were first discovered in baculoviruses because of their capacity to prevent virus-induced apoptosis and thereby facilitate virus multiplication (4,8). The baculovirus IAPs bear a striking resemblance to the cellular IAPs carried by the host insects that they infect. Cellular IAPs are a highly conserved family of survival factors that regulate developmental and stress-induced apoptosis, as well as inflammation, the cell cycle, and other signaling processes (35,38,44). Importantly, misregulation or overexpression of IAPs is associated with neoplasia and tumor chemoresistance (24,49). The IAPs are defined by the presence of one or more ϳ80-residue baculovirus IAP repeat (BIR) domains. The BIRs consist of a conserved Zn 2ϩ -coordinating arrangement of Cys and His residues (CCHC) that interact with diverse proteins, including the cysteinyl aspartate-specific proteases called caspases that execute apoptosis (reviewed in 16 and 37). The antiapoptotic activity of some, but not all, IAPs is derived from their ability to bind and neutralize caspases (reviewed in 35 and 44). The BIRs also interact with proapoptotic factors that contain IAP binding motifs (IBMs). IBM-containing factors have the capacity to bind and dissociate the IAP-caspase complex, thereby liberating active caspases to execute apoptosis (16,35,36,48). Many IAPs, including viral IAPs, also possess a C-terminal RING domain, which is a Zn 2ϩ -coordinating motif with E3ubiquitin ligase activity, which can contribute to antiapoptotic activity (48).
The best-studied baculovirus IAP is Op-IAP3, which is encoded by Orgyia pseudotsugata nucleopolyhedrovirus. This small IAP (268 residues) contains two BIRs and a C-terminal RING (Fig. 1A). Both BIRs are required for Op-IAP3 antiapoptotic activity (19,50,53). Truncation of the Op-IAP3 RING creates a loss-of-function dominant inhibitor (19). Op-IAP3's capacity to form a complex with this RING-lacking (RINGless) dominant inhibitor and with itself suggests that oligomerization is necessary for IAP function. Upon overex-pression, Op-IAP3 blocks apoptosis triggered by diverse signals in cells from certain insects and mammals, suggesting that it acts through a conserved mechanism (7,11,15,33,51,54,56). In the baculovirus host moth Spodoptera frugiperda (Lepidoptera: Noctuidae), Op-IAP3 prevents apoptosis by blocking the activation of effector caspases (25,32,40). However, in contrast to host insect IAPs, Op-IAP3 fails to inhibit active caspases (45,51,54). Thus, the host cell target(s) and the mechanism by which they are neutralized by this viral IAP remain unclear.
Among the cellular IAPs, SfIAP from Spodoptera frugiperda is most closely related to viral Op-IAP3. SfIAP (Fig. 1A) is 42% identical to Op-IAP3, with a higher degree of amino acid identity localized to its two BIRs and C-terminal RING (20). As the principal IAP in Spodoptera, SfIAP suppresses a constitutive push toward apoptosis (34); ablation of SfIAP leads to immediate apoptosis of cultured Spodoptera cells. Upon overexpression, SfIAP also rescues the multiplication of apoptosisinducing baculoviruses and can prevent apoptosis in certain mammalian cell lines (20,26). In contrast to viral Op-IAP3, SfIAP can bind and inhibit caspases, including Spodoptera frugiperda caspase-1 (Sf-caspase-1) and human caspase-9 (20,45). Thus, despite their structural similarities, there exist fundamental differences in the biochemical activities of these two IAPs. Importantly, SfIAP fails to prevent baculovirus-induced apoptosis when produced at endogenous levels in permissive Spodoptera cells. Thus, it is expected that SfIAP also possesses regulatory motifs that respond to cellular signals triggered upon virus infection.
SfIAP provides an unprecedented opportunity to investigate the functional and evolutionary relationships between host and viral IAPs and to test the intriguing hypothesis that viral IAPs were acquired by host gene capture (21). We have investigated the biochemical properties of SfIAP as a means to define its molecular mechanisms and to test its relatedness to viral IAPs. We report here that SfIAP shares many biochemical and functional features with viral IAPs. Like Op-IAP3, overexpressed SfIAP prevented virus-induced apoptosis at a step upstream of caspase activation by a mechanism that required BIR1, BIR2, and the RING. SfIAP formed a complex with itself and with a RINGless dominant inhibitor, suggesting that oligomerization is also required for function of cellular IAPs. Unlike viral IAPs, SfIAP possesses an N-terminal leader, which modulates intracellular SfIAP levels and may respond to apoptotic signals to regulate cell survival. Our data are consistent with a model in which baculoviruses acquired a host cell IAP and modified it for virus-specific needs, thereby increasing virus fitness by preventing virus-induced apoptosis.
Plasmids  IAPs. Viral Op-IAP3 (268 residues) and SfIAP (377 residues) each contain two BIR motifs (black boxes) and an E3 ligase RING domain (cross-hatched box). Each representing a potential start site, four methionines (M1 to M4) exist in the N-terminal leader of SfIAP. (B) SfIAP M4 mutations. SfIAP M4 (281 residues) begins with the M4 methionine. SfIAP M4 ⌬R (227 residues) lacks the C-terminal RING. Amino acid substitutions of Zn-coordinating residues are indicated. An epitope tag (HA) was inserted at the N terminus. (C) Marker rescue assay. The antiapoptotic activity of wild-type or mutated forms of SfIAP M4 was assayed by virus marker rescue in which replication of p35-deficient v⌬p35/lacZ was restored in proportion to the antiapoptotic activity of the mutated Sfiap gene acquired by integration of the SfIAP-encoding plasmid (2). Virus yields were determined by plaque assay using apoptosis-sensitive SF21 cells. Antiapoptotic activity is reported as the ratio of nonapoptotic, lacZ-expressing plaques produced by transfection of the indicated Sfiap to those produced by wild-type Sfiap. Values shown are the averages Ϯ standard deviations obtained from triplicate transfections.
Cell survival assays. SF21 monolayers (10 6 cells per 60-mm plate) were inoculated with the indicated AcMNPV recombinants. Photographs were taken of four evenly distributed fields of view from each of three replicate plates by using a Zeiss Axiovert 135TV microscope as described previously (19). Intact cells were counted, and percent survival is reported as the average ratios Ϯ standard deviations of intact, nonapoptotic cells. Values were normalized by adjusting the survival of vector-transfected cells infected with wild-type AcMNPV (p35 ϩ ) to 100%.
IAP dominant inhibition assays. For virus-based assays, triplicate plates of SF21 cells were transfected with plasmid and inoculated 24 h later with rvSfIAP HA (multiplicity of infection [MOI] ϭ 1 to 2 PFU per cell). Cells were collected 48 h later, lysed, and assayed for ␤-galactosidase by using the Galacto-Light Plus ␤-galactosidase chemiluminescent reporter assay (Tropix), as per the manufacturer's instructions. For UV irradiation-based assays, triplicate plates of 3 ϫ 10 6 DL-1 cells were transfected with the indicated lacZ reporter plasmids and irradiated 24 h later with UV-B (2 min) by using a Blak-Ray lamp (UVP, Upland, CA). Cells and associated apoptotic bodies were collected 24 h later, lysed, and assayed for ␤-galactosidase. All values are reported as the average activities Ϯ standard deviations obtained with triplicate plates.
Protein half-life measurements. SF21 cells transfected 24 h earlier with pIE1 prm -based plasmids encoding the indicated SfIAPs were overlaid with medium containing 400 g cycloheximide per ml; this drug concentration was the minimum sufficient to block Ͼ95% of SfIAP synthesis over a 4-h period (data not shown). Serial dilutions of cell lysates prepared at the indicated times thereafter were analyzed by using anti-T7 on duplicate immunoblots. SfIAP signal intensity was quantified by densitometry (NIH ImageJ) using films exposed within the linear range. The percent protein remaining at each interval after cycloheximide treatment was calculated and plotted as a function of time to generate protein half-lives (t 1/2 ) by best-fit analysis. Values are presented as the average values Ϯ standard deviations of protein half-lives (in minutes) derived from triplicate experiments.

SfIAP blocks apoptosis by a RING-dependent mechanism.
Spodoptera frugiperda SfIAP possesses two BIRs and a C-terminal RING motif and thus closely resembles baculovirus IAPs, including best-studied Op-IAP3 (Fig. 1A). To define biochemical differences between cellular and viral homologs, we first identified the required functional motifs within SfIAP.
Site-directed mutations were tested by plasmid transfection of Spodoptera frugiperda SF21 cells, which are highly sensitive to apoptotic stimuli and also permissive for baculovirus AcMNPV. The largest ORF within the Sfiap cDNA (20) is predicted to encode a 377-residue polypeptide (designated SfIAP M1 ) when initiated from the first available methionine (Fig. 1A). However, three other possible start codons are located N terminally to BIR1. Because it most closely resembles the size and structural organization of Op-IAP3, we first tested SfIAP M4 , which initiates from the fourth methionine (Fig. 1B). When expressed from transfected plasmids using the strong AcMNPV ie-1 promoter, HA epitope-tagged SfIAP M4 was synthesized as a single 281-residue species that was abundant and fully functional in SF21 cells (see below).
To test the antiapoptotic activity of SfIAP M4 , we transfected SF21 cells with SfIAP-encoding vectors and monitored cell survival after infection with the AcMNPV mutant v⌬p35, which lacks caspase inhibitor P35 and causes widespread apoptosis (18). Whereas v⌬p35 caused Ͼ75% apoptotic lysis of cells transfected with empty vector ( Fig. 2A), overexpressed SfIAP M4 prevented apoptosis. The survival of SfIAP M4 -transfected cells was more than 3-fold higher than that of control cells and comparable to that of cells infected with wild-type p35 ϩ AcMNPV (Fig. 2B). In contrast, deletion of the C-terminal RING motif caused a loss of antiapoptotic activity (Fig.  2B); this truncation (SfIAP M4 ⌬R) was readily produced in transfected cells (see below), which indicated that loss of function was not due to loss of protein. SfIAP M4 ⌬R loss of function was confirmed by marker rescue assays in which transfection with SfIAP M4 ⌬R-encoding plasmid failed to restore multiplication of the p35-null mutant v⌬p35/lacZ (Fig. 1C). In contrast, wild-type SfIAP M4 -encoding plasmid increased v⌬p35/lacZ multiplication by Ͼ500-fold. SfIAP M4 , but not SfIAP M4 ⌬R, also protected cells from UV-induced apoptosis (see below). Thus, overexpressed SfIAP prevented apoptosis triggered by diverse signals in a RING-dependent manner, a finding consistent with earlier studies (20).
SfIAP blocks effector caspase activation. Apoptosis in Spodoptera involves sequential activation of initiator and effector caspases, which includes the principal effector Sf-caspase-1 (1,25,31). Activated first in the caspase pathway, initiator caspases proteolytically cleave and thereby activate effector caspases from their inactive proforms (16,36,37). Viral Op-IAP3 blocks apoptosis by preventing Sf-caspase-1 activation (25,40,56). To define the step at which SfIAP acts, we monitored the proteolytic processing of pro-Sf-caspase-1, which is cleaved to large and small subunits during activation (1,25). To ensure that SfIAP M4 was overproduced in every cell trig-gered to undergo apoptosis, we generated an AcMNPV recombinant vSfIAP, which lacks p35 but expresses Sfiap M4 from the early ie-1 promoter. Upon infection with vSfIAP, processing of pro-Sf-caspase-1 was blocked (Fig. 2C, lane 1); the abundance of SfIAP M4 (see below) indicated that infection proceeded normally. In contrast, pro-Sf-caspase-1 was processed to active subunits upon infection with p35 ϩ recombinant vP35 (Fig. 2C, lane 2); P35 is a substrate inhibitor of effector caspases that blocks virus-induced apoptosis by inhibiting activated Sfcaspase-1 (56). We concluded that SfIAP M4 acts upstream of P35 and thus prior to effector caspase activation.
SfIAP M4 inhibition of pro-Sf-caspase-1 processing was comparable to that by P49 and Op-IAP3, produced by AcMNPV recombinants vP49 and vOpIAP, respectively (Fig. 2C, lanes 4  to 7). Bearing considerable sequence similarity to P35, baculovirus P49 is a substrate inhibitor that is cleaved by and inhibits the initiator caspase that processes Sf-caspase-1 (56). Thus, to determine if SfIAP also inhibits the P49-sensitive initiator caspase, we tested whether SfIAP M4 affects P49 cleavage. When P49 and SfIAP M4 were expressed simultaneously in cells coinfected with recombinants vP49 and vSfIAP, cleavage of P49 was blocked (Fig. 2D). P49 cleavage was detected only in the absence of vSfIAP (Fig. 2D, lane 6), indicating that SfIAP M4 inhibition was dominant to that by P49. We concluded that when overproduced, SfIAP M4 inhibits the activation or activity of the Spodoptera initiator caspase responsible for virus-induced apoptosis. Thus, SfIAP can function at a step also inhibited by viral Op-IAP3 (56). RINGless SfIAP is a dominant inhibitor of SfIAP. Viral Op-IAP3 RING truncations are nonfunctional and potently inhibit wild-type Op-IAP3 (19). To determine if SfIAP behaves similarly, we tested the effect of SfIAP M4 RING mutations on SfIAP function. To this end, we engineered a p35-null AcMNPV recombinant (rvSfIAP/lacZ) that produces SfIAP M4 at levels minimally sufficient to block apoptosis (see below). rvSfIAP/lacZ includes a polyhedrin promoter-directed lacZ reporter, which is expressed only when apoptosis is suppressed because the polyhedrin promoter is not activated until very late in infection (18). Thus, ␤-galactosidase production provided a sensitive measure of apoptotic inhibition, as used previously for studies of Op-IAP3 inhibitors (19). SfIAP M4 produced by rvSfIAP/lacZ blocked virus-induced apoptosis (Fig. 3A); cell survival 48 h after infection was Ͼ95%. In contrast, SfIAP M4 -mediated inhibition of virus-induced apoptosis was reduced by prior transfection with plasmid encoding RINGless SfIAP M4 ⌬R; apoptotic lysis affected Ͼ90% of the cells (Fig. 3A). SfIAP M4 ⌬R increased apoptosis of rvSfIAP/lacZ-infected cells in a dose-dependent manner, as indicated by decreasing ␤-galactosidase levels with increasing amounts of plasmid (Fig. 3B). Because transfection efficiencies were Ͻ100%, ␤-galactosidase activity was not reduced to zero. Immunoblots indicated that SfIAP M4 ⌬R was more abundant than virus-expressed SfIAP M4 in these assays (data not shown). These data suggested that SfIAP M4 ⌬R dominantly interferes with SfIAP.
To confirm this conclusion, we used heterologous Drosophila DL-1 cells that lack endogenous SfIAP and are sensitive to UV radiation-induced apoptosis. DL-1 cells were transfected with plasmids encoding either SfIAP M4 or RINGless SfIAP M4 ⌬R and cotransfected with plasmid pMT-lacZ/pIE1-Sfiap M4 that encodes ␤-galactosidase under the control of the Drosophila metallothionine promoter and SfIAP M4 under the control of the ie-1 promoter (Fig. 3C). After UV irradiation, ␤-galactosidase production was used to monitor cell viability. Cell survival was highest when only wild-type SfIAP M4 was expressed (Fig. 3D), indicating that SfIAP M4 blocks UV-induced apoptosis. In contrast, cell survival was lowest upon coexpression of SfIAP M4 ⌬R and SfIAP M4 ; ␤-galactosidase was reduced by 4to 5-fold compared to that of cells transfected with vector alone (Fig. 3D). We concluded that SfIAP M4 ⌬R dominantly inhibits the antiapoptotic activity of SfIAP in heterologous and homologous cells.
SfIAP⌬R interacts with SfIAP. To test whether dominant inhibition involved decreased SfIAP stability, we monitored intracellular protein levels. Immunoblot analysis indicated that SfIAP M4 ⌬R did not affect SfIAP M4 levels either in plasmidtransfected cells (Fig. 4A, lanes 1 to 3) or in vSfIAP-infected cells (lanes 5 to 7). Consequently, we determined whether SfIAP M4 ⌬R interacts with SfIAP M4 . In immunoprecipitation assays, HA-tagged SfIAP M4 ⌬R was precipitated with T7tagged SfIAP M4 (Fig. 4B, lane 2). Little, if any, SfIAP M4 ⌬R precipitated nonspecifically (Fig. 4B, lane 5). These interactions were confirmed by immunoprecipitations in which the After 24 h, the cells were inoculated with rvSfIAP/lacZ, and intracellular ␤-galactosidase was measured 48 h later. In this assay, the yield of ␤-galactosidase is directly proportional to the level of apoptotic protection, which is reported as the average value Ϯ standard deviation of activity relative to that of vector-transfected cells (normalized to 100%) obtained from triplicate infections. (C) Expression vectors. The ie-1 or metallothionine (MT) promoters (prm) were linked to Sfiap M4 or lacZ, respectively, and inserted into the indicated plasmids. An arrow denotes the RNA start sites. (D) Antiapoptotic activity of SfIAP M4 in Drosophila cells. DL-1 cells were transfected with empty vector or plasmids encoding the indicated proteins, UV irradiated 24 h later, and immediately overlaid with Cu 2ϩ -containing medium to induce lacZ expression. The level of apoptotic protection was directly proportional to the yield of 24-h intracellular ␤-galactosidase, which is reported as the average value Ϯ standard deviation of activity relative to that of cells cotransfected with empty vector and pMT-lacZ/pIE1-Sfiap (normalized to 100%) obtained from triplicate transfections. epitope tags were reversed (data not shown). Thus, SfIAP⌬R forms a stable complex with SfIAP. Moreover, full-length HAtagged SfIAP M4 coprecipitated with full-length T7-tagged SfIAP M4 (Fig. 4B, lane 1); SfIAP M4 was not precipitated in the absence of SfIAP T7 (Fig. 4B, lane 4). We concluded that SfIAP also interacts with itself and forms homo-oligomeric complexes like viral Op-IAP3 (19).
Intracellular SfIAP is dimeric. The capacity of SfIAP to interact with itself suggested that it forms oligomers. To clarify the nature of homo-oligomerization, we used size exclusion chromatography to characterize the IAP complex. The majority of SfIAP M4 obtained from transfected SF21 cells by freezethaw lysis eluted as a complex, with a molecular mass of 60 to 70 kDa (Fig. 5A). Since the monomeric mass of SfIAP M4 (281 residues) is 32 kDa, we concluded that the principal form of SfIAP is a ϳ64-kDa dimer under these conditions. Minor amounts of SfIAP were reproducibly detected in fractions ex-pected for larger complexes, suggesting the possibility of interaction with other cellular proteins. Collectively, our data argue that a majority of overexpressed SfIAP exists as a homodimer rather than in a complex with other apoptosis-associated molecules. When overexpressed in SF21 cells, 30-kDa Op-IAP3 (268 residues) also eluted with a size expected of a dimeric complex (Fig. 5B), consistent with its capacity to homooligomerize (19). Op-IAP3 had a greater tendency to form larger complexes, possibly through multimerization or interaction with cellular components.
RINGless SfIAP triggers caspase activation. Because SF21 cells exhibit a constitutive push toward caspase activation (34), we predicted that SfIAP M4 ⌬R-mediated inhibition of endogenous SfIAP would trigger caspase activity. To test this possibility, we determined the effect of SfIAP M4 ⌬R on caspase activation in SF21 cells. In contrast to transfection of vector or SfIAP M4 -encoding plasmid (Fig. 6A, lanes 1 to 3 and 4 to 6), transfection with SfIAP M4 ⌬R-encoding plasmid triggered the processing of Sf-caspase-1 to active subunits (lanes 7 to 9). SfIAP M4 ⌬R-induced processing was comparable to that induced by overexpression of Drosophila proapoptotic factor HID (Fig. 6A, lanes 19 to 21), which causes widespread apoptosis of SF21 cells (50). The intracellular level of SfIAP M4 ⌬Rinduced caspase activity was 3-fold higher than that of vectortransfected cells and comparable to that induced by HID when measured using the effector caspase substrate DEVD-AMC (Fig. 6B). As expected, caspase activity was lowest in SfIAP M4expressing cells (Fig. 6B). Immunoblots readily detected transfected SfIAP M4 ⌬R, which was more abundant than SfIAP M4 (Fig. 6A, lanes 4 to 6 and 7 to 9). We concluded that SfIAP M4 ⌬R triggers caspase activation and thus mimics RINGless Op-IAP3 (19).

BIR1 of SfIAP⌬R is required for dominant inhibition.
To define the domain(s) responsible for inhibition by SfIAP M4 ⌬R, we first examined the BIRs, which interact with known proapoptotic ligands. Substitutions of the Zn-coordinating Cys and His residues (Fig. 1B) were used to disrupt the folding of each BIR. When tested individually by marker rescue assays,  7) and lysed 24 h later. All lysates were subjected to immunoblot analysis using anti-HA (top) and anti-actin (bottom). (B) Immunoprecipitations. SF21 cells were transfected (ϩ) with empty vector or plasmid encoding the indicated T7-or HA-tagged proteins, and NP-40-treated extracts were prepared 24 h later. Proteins were immunoprecipitated (ppt) by using anti-T7 beads and subjected to immunoblot analysis (blot) by using anti-T7 or anti-HA. Lysates obtained prior to immunoprecipitation (total lysate) were included.  (Fig. 1C), indicating that both BIRs are required for antiapoptotic activity. These BIR substitutions were then introduced into SfIAP M4 ⌬R for tests of dominant inhibition in our rvSfIAP/lacZ ␤-galactosidase assay. Upon transfection, each BIR-mutated SfIAP M4 ⌬R was readily detected (see below). BIR2 substitutions C154S and C178S had no effect on dominant inhibition by SfIAP M4 ⌬R, since both proteins reduced antiapoptotic activity of virus-encoded SfIAP M4 just as well as nonmutated SfIAP M4 ⌬R (Fig. 7A). Conversely, substitutions C43A, H60A, and C67A within BIR1 caused loss of dominant inhibition by SfIAP M4 ⌬R (Fig. 7B). C67A-and C178S-doubly mutated SfIAP M4 ⌬R also failed to inhibit SfIAP M4 (data not shown). Thus, only BIR1 is required for dominant inhibition. We confirmed these findings by showing that BIR1 is also required for SfIAP M4 ⌬R-mediated inhibition of endogenous cellular SfIAP. Processing of Sf-caspase-1 to active subunits was blocked or reduced in the presence of BIR1 C43A-or C67A-mutated SfIAP M4 ⌬R (Fig. 6A, lanes 10 to 15). Conversely, processing in the presence of BIR2 C178S-mutated SfIAP M4 ⌬R (Fig. 6A, lanes 16 to 18) was comparable to that of dominant inhibitor SfIAP M4 ⌬R and HID. Intracellular caspase activity was also reduced in the presence of BIR1 C43A-or C67A-mutated SfIAP M4 ⌬R but was significantly increased by BIR2 C178S-mutated SfIAP M4 ⌬R (Fig. 6B), indicating that BIR1, not BIR2, was required for SfIAP M4 ⌬Rmediated inhibition.

BIR-disrupted SfIAP⌬R interacts with SfIAP.
To determine whether BIR disruption interfered with SfIAP M4 ⌬R interaction with functional SfIAP, we used immunoprecipitation assays. Full-length T7-tagged SfIAP M4 formed a complex with BIR2 C154S-and C178S-mutated SfIAP M4 ⌬R (Fig. 8A, lanes 4  and 6) and with BIR1 C43A-, H60A-, and C67A-mutated SfIAP M4 ⌬R (Fig. 8B, lanes 4, 6, and 8). These interactions were as strong as or stronger than those with unaltered SfIAP M4 ⌬R ( Fig. 8A and B, lane 2). Full-length T7-tagged SfIAP M4 also interacted with C67A-and C178S-mutated SfIAP M4 , in which both BIRs were mutated (data not shown). None of the SfIAP M4 mutations were precipitated in the absence of full-length SfIAP T7 (Fig. 8), and all interactions were confirmed by reciprocal assays (data not shown). We concluded that the proper folding of BIR1 or BIR2 is not required for interaction between SfIAP M4 ⌬R and SfIAP. These findings suggested that a domain(s) outside or independent of the BIRs mediates SfIAP homophilic interactions.

BIR disruptions within full-length SfIAP generate dominant inhibitors.
To further define the role of the BIRs in dominant inhibition, we tested the effect of BIR disruptions within fulllength SfIAP. When expressed by plasmid transfection, BIR1 C43A-, H60A-, and C67A-mutated SfIAP M4 and BIR2 C154Sand C178S-mutated SfIAP M4 dominantly interfered with the antiapoptotic activity of SfIAP M4 (Fig. 7A and B); each of these mutated RING-containing proteins reduced the yield of ␤-galactosidase within rvSfIAP/lacZ-infected cells, indicating that SfIAP antiapoptotic activity was compromised. Dominant inhibition by these BIR-mutated proteins was comparable to that of RINGless SfIAP M4 ⌬R. In contrast, C67A-and C178Smutated SfIAP M4 , in which both BIRs were disrupted, failed to alter SfIAP M4 function in this assay (Fig. 7A). This doubly mutated SfIAP M4 was readily produced in transfected cells (data not shown), ruling out the possibility that protein instability caused the loss of dominant inhibition. We concluded that disruption of either BIR1 or BIR2, but not both, can generate a dominant inhibitor of functional SfIAP M4 .
As demonstrated by immunoprecipitations, BIR2 C154Sand C178S-mutated SfIAP M4 formed a complex with fulllength T7-tagged SfIAP M4 (Fig. 8A, lanes 3 and 5). Likewise, BIR1 C43A-, H60A-, and C67A-mutated SfIAP M4 coprecipitated with full-length SfIAP M4 (Fig. 8B, lanes 3, 5, and 7). Moreover, C67A-and C178S-doubly mutated SfIAP M4 also interacted with full-length SfIAP (data not shown). These findings confirmed our conclusion that although at least one BIR is required for dominant inhibition, properly folded BIRs are not required for SfIAP complex formation. The finding that C67A-and C178S-mutated SfIAP M4 also formed a complex with SfIAP but failed to act as a dominant inhibitor suggests that interaction with a nonfunctional IAP is necessary but not sufficient for dominant inhibition. Additional studies are required to define the mechanisms involved.
The N-terminal SfIAP leader regulates protein abundance. A striking feature of invertebrate (lepidopteran) host IAPs is the possession of a long N-terminal leader sequence that precedes BIR1 (Fig. 1A). The absence of a comparable sequence in viral IAPs (20,45) suggests that the leader performs a function specific for cellular IAPs. To investigate the potential role of the SfIAP leader, we generated an expression vector for SfIAP M1 , which uses the first available methionine (M1) in the Sfiap ORF and thus includes the 100 residues prior to BIR1 (Fig. 9A). SfIAP M1 was engineered to include an epitope (HA) tag. Immunoblotting of transfected SF21 cells detected fulllength SfIAP M1 (ϳ54 kDa) but at relatively low steady-state levels compared to that of SfIAP M4 (Fig. 9B and C, compare  lanes 3 and 9). Smaller SfIAP M1 -related products were recognized by anti-SfIAP but not by anti-HA. Because these products lacked the N-terminal HA tag, they were derived by either cleavage or internal initiation from the second or third methionine (M2 or M3) within the leader (Fig. 9A). Deletion of the RING (SfIAP⌬R M1 ) or disruption of the E3 ubiquitin ligase activity of the RING (51) by the substitution H347A caused a modest increase in SfIAP M1 steady-state levels ( Fig. 9B and C,  7). Nonetheless, leaderless versions of SfIAP M4 (Fig. 9B, lanes 9, 11, 13, and 15) were routinely detected at significantly higher levels than in wild-type or RING-mutated forms of SfIAP M1 . The SfIAP leader confers protein instability. To test whether the higher accumulation of leaderless SfIAP M4 was due to increased stability, we measured the protein half-lives. T7 epitope-tagged SfIAP M1 and SfIAP M4 were overproduced in SF21 cells by using our ie-1 promoter-based expression vector (Fig. 10A). Because both proteins were translated from the same AUG start codon within the same 5Ј noncoding mRNA leader, the transcriptional and translational contexts of each construct were identical. Measurements of protein stability were made after protein synthesis was blocked by cycloheximide. We discovered that leader-containing SfIAP M1 was turned over ϳ5 times faster than SfIAP M4 (Fig. 10B). Whereas the 33-min half-life of SfIAP M1 was comparable to that of Drosophila IAP1 (DIAP1; t 1/2 ϭ 30 to 40 min) (52,55), deletion of the leader increased the half-life of SfIAP M4 to 152 min. Moreover, the longer half-life of SfIAP M4 was comparable to that of I332A-mutated SfIAP M1 (Fig. 10B), which is deficient for E3 ubiquitin ligase activity of the RING (5,17,30). The T7 epitope tag at the N terminus had no significant effect on stability of these proteins (data not shown). We concluded that the N-terminal leader of SfIAP decreases protein stability and that both the leader and the RING contribute to SfIAP turnover under steady-state conditions. Steady-state levels of SfIAP M1 are unaffected by caspases. The SfIAP leader possesses three potential caspase cleavage sites at Asp83, Asp86, and Asp96, immediately N terminal to BIR1 (Fig. 9A). One of these sites, DKTD 83 2N, has been implicated in SfIAP turnover and anti-caspase activity because cleavage here exposes a domain required for N-end rule ubiquitination and degradation (45). To assess the contribution of caspase-mediated cleavage to SfIAP M1 instability, we first determined the effect of caspase inhibitors on SfIAP levels. When effector caspase inhibitor P35 was coexpressed (Fig. 9D), caspase activity as measured by DEVD-AMC cleavage was reduced to background levels in cells producing dominant inhibitors of endogenous SfIAP (Fig. 9E). However, this P35 anti-caspase activity had little if any effect on the steady-state levels of SfIAP M1 (Fig. 9B and C, compare lanes 3 and 4). The level of RINGless SfIAP⌬R M1 and RING-defective H347Amutated SfIAP M1 increased modestly (Fig. 9B, compare lanes 5 and 6 to lanes 7 and 8). We attributed this small difference to P35 inhibition of caspases activated by the dominant inhibitory activity of these proteins (Fig. 9E). Consistent with caspase cleavage within the leader, SfIAP⌬R M1 and H347A-mutated SfIAP M1 were accompanied by smaller fragments (Fig. 9C, asterisks) that disappeared upon P35 coexpression (Fig. 9C, lanes 5 to 8). P35 had little or no effect on SfIAP M4 (Fig. 9B, lanes 9 to 16), suggesting that leaderless SfIAP is insensitive to caspase cleavage.
To rule out the possibility that initiator caspase activity affected SfIAP stability, we also coexpressed baculovirus P49, which is a direct inhibitor of the Spodoptera initiator caspase responsible for effector caspase activation (56). When compared directly with P35, P49 had no effect on the steady-state levels of leader-containing SfIAP M1 (Fig. 9F, lanes 1 to 3 and 7 to 9). Like P35, P49 prevented the appearance of the caspase cleavage products of dominant inhibitors SfIAP⌬R M1 and H347A-mutated SfIAP M1 (Fig. 9F, compare lanes 7 and 10 with lanes 9 and 12). Thus, neither initiator nor effector caspase activity affected the levels of wild-type SfIAP. Confirming this conclusion, the steady-state level of caspase cleavage-resistant D83A-mutated SfIAP M1 was comparable to that of wild-type SfIAP M1 (Fig. 9F, compare lanes 1 to 3 with lanes 4 to 6). Collectively, these data indicated that caspase activity fails to explain the reduced steady-state levels of SfIAP and suggested that the N-terminal leader regulates protein turnover by a caspase-independent mechanism.

DISCUSSION
SfIAP is the principal regulator of apoptosis in the baculovirus host Spodoptera frugiperda. We report here that on the basis of intracellular activities and biochemical properties, SfIAP closely resembles baculovirus Op-IAP3. Upon overexpression, both IAPs require the same motifs and prevent apoptosis at a step upstream of caspase activation. Nonetheless, endogenous SfIAP is a low-abundance protein that is rapidly depleted upon infection and thus fails to protect cells from virus-induced apoptosis (R. J. Cerio, K. Schultz, R. Vandergaast, and P. D. Friesen, unpublished data). As shown here, this intracellular instability of SfIAP is governed in part by its N-terminal leader sequence, which is absent within viral IAPs. These differences and similarities between cellular and viral IAPs have provided important insight into the mechanisms of IAP function in invertebrates and are consistent with the hypothesis that baculovirus IAPs were derived by capture of host cell IAP (21).

SfIAP inhibition of caspase activation and activity.
Sitedirected mutagenesis demonstrated that SfIAP requires BIR1, BIR2, and its C-terminal RING motif for antiapoptotic activity when overexpressed in Spodoptera cells ( Fig. 1 and 2). Given that SfIAP has high (42%) sequence similarity to baculovirusencoded Op-IAP3 (20), the critical nature of these defining motifs was predictable. For certain cellular IAPs like DIAP1, the BIRs bind and suppress the activity of caspases (36). Indeed, SfIAP can interact with effector caspases through its BIRs (45; R. Vandergaast, D. Tran, and P. D. Friesen, unpublished data). However, when overexpressed, SfIAP functioned upstream of effector caspase activation, as SfIAP blocked the proteolytic cleavage of pro-Sf-caspase-1 in Spodoptera cells (Fig. 2D) and pro-DrICE in Drosophila cells (26). Furthermore, overexpressed SfIAP blocked the cleavage of baculovirus P49 (Fig. 2C), which is a direct substrate inhibitor of the initiator caspase that activates Sf-caspase-1 (25,56). Thus, SfIAP can function at or upstream of the activation of the P49-sensitive initiator caspase that mediates baculovirus-induced apoptosis. As such, SfIAP probably inhibits the same step in the apoptosis pathway as baculovirus Op-IAP3 or a step adjacent to that inhibited by Op-IAP3 (56). Nonetheless, whereas SfIAP has the capacity to bind active effector caspases and block apoptosis in lepidopteran and dipteran cells, Op-IAP3 does not (26,45,46,54). Thus, there exist significant differences in the core mechanisms by which SfIAP and viral IAPs block apoptosis.
Regulation of SfIAP stability. Endogenous SfIAP is a lowabundance protein that is difficult to detect when directly compared to those produced by our plasmid expression vectors (Fig. 9B). However, by using highly purified antiserum and Sfiap-specific RNA silencing, we have determined that SfIAP M1 is the principal translation product of Sfiap in SF21 cells (R. Vandergaast, R. J. Cerio, and P. D. Friesen, unpublished data). When overproduced at equivalent levels, SfIAP M1 and SfIAP M4 have comparable antiapoptotic activity levels. Nonetheless, we demonstrated here that SfIAP M1 is significantly less stable than SfIAP M4 ( Fig. 9 and 10). Full- length SfIAP M1 exhibited a half-life of ϳ30 min, which is comparable to that of DIAP1, the principal IAP of Drosophila. However, removal of the 99-residue SfIAP leader increased protein stability by a factor of 5 and indicated that the Nterminal leader confers significant instability to SfIAP M1 . Thus, although the N-terminal leader does not directly affect antiapoptotic function, it is a principal regulator of intracellular levels of SfIAP. N-end rule ubiquitination and degradation has been implicated in the regulation of DIAP1 levels (10,45). Caspasemediated cleavage of the N-terminal DIAP1 leader at DQVD 20 2N exposes an N-terminal Asn residue, thereby recruiting N-end rule ubiquitination machinery, which destabilizes DIAP1. Although caspase cleavage sites are located in the N-terminal leader of SfIAP, including DKTD 83 -N (Fig. 9A), our studies here suggested that caspase activity has a minimal effect on SfIAP stability. In particular, the levels of wild-type SfIAP M1 were unaffected when caspase activity was inhibited by coexpression of baculovirus caspase inhibitors P35 and P49 (Fig. 9). Confirming this conclusion, the intracellular level of caspase-resistant D83A-mutated SfIAP M1 was comparable to that of wild-type SfIAP M1 (Fig. 9F). Importantly, our study also demonstrated that leaderless SfIAP M4 , which lacks all N-terminal caspase-cleavage sites, exhibited potent antiapoptotic activity in Spodoptera and Drosophila cell lines ( Fig. 2 and  3) (26). These findings suggest that caspase cleavage is not required for SfIAP antiapoptotic function.
Our study also suggests that SfIAP motifs other than that of the N-terminal leader regulate protein stability. For example, disruption of BIR1 also reduced steady-state levels of leaderless SfIAP M4 (Fig. 8B). Inactivation of the C-terminal RING restored protein levels, suggesting that BIR1 and the predicted E3 ubiquitin ligase activity of the RING contribute to SfIAP stability. Loss of its RING also stabilizes DIAP1 (52). The observation that SfIAP possesses multiple determinants that affect its intracellular concentration is consistent with the critical role that this cellular IAP plays in determining cell fate, including that after virus infection.
Role of oligomerization in IAP function. Immunoprecipitations and size exclusion chromatography ( Fig. 4 and 5) indicated that SfIAP M4 exists primarily as a homodimer. The finding that RINGless SfIAP M4 forms a complex with and inhibits wild-type SfIAP (Fig. 3, 4, and 6) also suggested that oligomerization contributes to antiapoptotic function. As such, SfIAP mimics viral Op-IAP3, which also forms homodimers (Fig. 5) and is potently inhibited by RINGless mutations (19). Homoand hetero-oligomerization have been described for human IAPs, including XIAP and c-IAP1. The domains required for IAP oligomerization vary (28,29,42,43). In the case of Op-IAP3, the RING is dispensable for oligomerization (19). Likewise, the SfIAP RING and BIRs are not required for homo-oligomeric interactions (Fig. 7). Nonetheless, other experiments have indicated that the isolated SfIAP RING interacts with SfIAP (D. Tran and P. D. Friesen, unpublished data). Thus, it is likely that multiple domains contribute to SfIAP oligomerization.
The biological relevance of IAP oligomerization is unknown. Our findings here that RINGless SfIAP interacts and interferes with wild-type SfIAP suggest that two RING motifs are required for E3 ligase ubiquitination activity. Dimerized RING motifs confer optimal E3 activity from certain RING-contain-ing proteins, including human XIAP (9,14,23,43). SfIAP may require cooperative interaction between two RING motifs as well. Thus, a complex of SfIAP⌬R with full-length SfIAP might have insufficient ubiquitination activity, which would increase levels of caspase normally turned over by interaction with endogenous SfIAP. Consistent with this possibility, SfIAP⌬R increased caspase activation and activity in transfected cells without affecting the level of SfIAP (Fig. 4 and 6). SfIAP dimerization may also be necessary for proper binding of ligands destined for ubiquitination and turnover. Because caspases form active dimers (37), a dimeric BIR may be necessary to provide a stable interface for RING-mediated inactivation by ubiquitination. Likewise, a dimeric BIR interface may be necessary for interaction with IBM-containing prodeath factors. Our studies indicate that the oligomeric state of an IAP is an important consideration for mechanistic studies.
Dominant inhibition by loss-of-function IAPs. Disruptions of either the RING or the BIR motifs of SfIAP functioned as dominant negative mutations (Fig. 3, 6, and 7). These findings suggest a common mode of action for cellular and viral IAPs. In contrast to RINGless Op-IAP3 (19), the dominant inhibitory, proapoptotic activity of RINGless SfIAP was selectively lost upon disruption of BIR1 ( Fig. 6 and 7); BIR2 was dispensable for this activity. This finding implies that BIR1 interacts with a factor(s) involved in dominant inhibition and that the function of BIR1 is linked to that of the RING. It has been reported that BIR1-containing SfIAP truncations interact with active effector Sf-caspase-1 (45). Thus, one of several possible mechanisms for dominant inhibition by RINGless SfIAP involves the binding of caspases to BIR1 in such a way that caspase activity is not destroyed by ubiquitination. This nonproductive titration would increase intracellular caspase levels ( Fig. 6) and sensitize the cell to apoptosis. Disruptions of BIR1 or BIR2 within full-length SfIAP also generated dominant inhibitors (Fig. 7). This finding is of particular interest because it indicates that a dominant inhibitor can possess a RING motif. These full-length mutations also complexed with SfIAP ( Fig. 9), suggesting that their inhibitory effects involved direct interaction with SfIAP. Interestingly, BIR1-but not BIR2disrupted SfIAP was less stable with rather than without its RING (Fig. 8). This finding suggested that BIR1-disrupted SfIAP may be more sensitive to autoubiquitination and that BIR1 regulates RING-mediated ubiquitination. Additional experiments are required to test this important possibility.
Host origin for viral IAPs? To date, there are few unequivocal cases of direct host gene capture (transfer) by DNA viruses (12,21,22). On the basis of phylogenetic relatedness, it has been proposed that baculovirus-encoded IAPs were obtained through the capture of host lepidopteran IAPs on at least two independent occasions (21). Assuming that this acquisition was relatively recent, viral and cellular IAPs should share structural motifs, biochemical properties, and interactions with host components. As we have shown here, SfIAP requires the same functional motifs (BIR1, BIR2, and RING) and acts at the same step to block caspase activation and apoptosis as baculovirus Op-IAP3. Moreover, RINGless truncations of both of these IAPs act as dominant inhibitors of wild-type IAP. Both IAPs form intracellular dimers and can interact with their dominant inhibitors. Interestingly, viral Op-IAP3 can also interact with SfIAP, indicating that oligomer-ization domains may be shared (R. J. Cerio and P. D. Friesen, unpublished data). Collectively, these data support a common origin for both proteins.
The most striking feature that distinguishes SfIAP is its N terminus. A 30-to 100-residue leader preceding BIR1 is a prominent attribute of many cellular IAPs of lepidopteran, dipteran, and hymenopteran species. However, most, if not all, viral IAPs either lack the leader or possess just a few residues prior to BIR1. Our studies indicate that the N-terminal SfIAP leader is a principal determinant of protein turnover. We expect that the leader possesses a regulatory motif(s) that provides a critical means by which to govern intracellular concentrations of the host IAP to meet cellular demands for preventing or promoting apoptosis. In the case of virus infection, virus-induced depletion of cellular IAPs would lead to rapid apoptosis, as observed with DIAP1 in RNA virus-infected Drosophila cells (41). In contrast, viral IAPs function over a narrow window of infection to prevent apoptosis and thus would benefit from immunity to the mechanisms downregulating cellular IAP levels. Thus, a virus-encoded IAP that lacks an N-terminal regulatory leader or is deficient in hostresponsive destabilizing motifs would have improved protein stability and longer-lasting antiapoptotic activity compared to those of host IAPs. Our study here suggests the simple possibility that viral IAPs evolved from cellular IAPs in a process that included the loss of the N-terminal regulatory leader. Additional studies promise to provide insight into this important aspect of host-virus evolution.