Sequence requirements for Hid binding and apoptosis regulation in the baculovirus inhibitor of apoptosis Op-IAP. Hid binds Op-IAP in a manner similar to Smac binding of XIAP.

It has been suggested that the Drosophila Hid protein interacts with the baculovirus Op-IAP protein in a manner similar to that of human Smac binding to XIAP, based largely on amino acid sequence homology. However, there is little direct experimental evidence in support of this hypothesis; indeed, evidence exists from previous studies suggesting that the mode of binding is not similar. We have now precisely mapped the interaction between Hid and Op-IAP, and we show clearly for the first time that the biochemical interactions between the amino terminus of Hid and BIR2 of Op-IAP are highly similar to those found between the processed amino terminus of Smac and BIR3 of XIAP. Also similar to Smac, the amino terminus of Hid must be processed to bind Op-IAP. In addition, our data also suggest that a second interaction between Hid and Op-IAP exists that does not involve the amino terminus of Hid, which may explain some of the earlier contradictory results. The evolutionary conservation of this mechanism of binding underscores its importance in apoptotic regulation. Nevertheless, interaction with Hid is not sufficient for Op-IAP to inhibit apoptosis induced by Hid overexpression or by treatment with actinomycin D, indicating that additional sequence elements are required for the anti-apoptotic function of Op-IAP.

The first inhibitor of apoptosis (iap) 1 genes were discovered in baculoviruses (1,2), and iap homologs have since been found in cellular genomes ranging from yeast to humans. Most iap genes from higher eukaryotes are involved in inhibiting apoptosis, whereas those from yeast and nematodes have roles in regulating cytokinesis (3). The hallmark motif of the IAP protein family is an ϳ70-amino acid domain termed the baculovirus IAP repeat (BIR), so named because 1-3 copies are found in the amino-terminal region of all IAP proteins. Although these domains are termed "repeats," they are not merely exact copies of each other, and different BIR domains have been shown to have different activities. In addition to the residues within the core BIR domain, the non-conserved sequences that flank BIRs have also been shown to be important for anti-apoptotic function. For example, the flanking sequences upstream of the second BIR domain of human XIAP directly bind and inhibit caspase-3, whereas the third BIR domain itself binds and inhibits caspase-9 (4 -6). Various BIR domains and/or their flanking sequences have been shown to mediate different protein-protein interactions, including self-oligomerization (7), binding, and inhibition of certain caspases (8), and binding to a number of other proteins including the Drosophila pro-apoptotic proteins Hid, Reaper, and Grim (9,10), as well as the mammalian pro-apoptotic protein Smac/DIABLO (11,12), among others.
Most cellular IAP proteins, such as DIAP1 from Drosophila and the human proteins XIAP, c-IAP1, and c-IAP2, appear to be able to block apoptosis by binding and inhibiting certain caspases directly, although compelling evidence for additional mechanisms also exists, including E3 ubiquitin-protein ligase activity and interaction in signal transduction pathways (reviewed in Refs. 13 and 14). Expression of the baculovirus Op-IAP protein has been shown to inhibit the processing of a caspase called Sf-caspase-1 in the lepidopteran cell line Sf21, a cell line derived from the fall armyworm, Spodoptera frugiperda (15). The processing of Sf-caspase-1 is presumably carried out by an as yet unidentified apical caspase in Sf21 cells that has been termed Sf-caspase-X (16). However, even though Op-IAP inhibits the processing of Sf-caspase-1, it has yet to be shown that a direct caspase interaction is responsible for the inhibitory action of Op-IAP. Genetic evidence from Drosophila supports a model wherein Hid, Reaper, and Grim induce apoptosis by binding to the IAP protein DIAP1 and displacing bound caspases, thereby initiating caspase activation (17,18). Similarly, the mammalian Smac protein appears to induce apoptosis by displacing caspase-9 from XIAP (19).
Recently, the crystal structure of the complex formed by Smac and the BIR3 domain of the human XIAP protein was determined (20,21). The amino terminus of Smac is processed upon translocation into mitochondria, revealing a sequence element with homology to the amino terminus of Hid, Reaper, and Grim. The first four amino acids of the processed Smac amino terminus bind in a groove within the core BIR3 domain of XIAP, whereas a separate region of Smac is also involved in an additional site of interaction with BIR3. The similarity of the first four amino acids of the processed amino terminus of Smac (Ala-Val-Pro-Ile) to amino acids 2-5 of Hid (Ala-Val-Pro-Phe) led to the suggestion that Hid may be post-translationally processed by a methionine aminopeptidase, and the amino terminus of Hid may bind to IAPs by a similar mechanism to that of Smac. In support of this hypothesis, a peptide lacking the initiating methionine of Hid bound to BIR3 of XIAP, but a peptide containing the initiating methionine did not bind (20). The recent description of the structure of BIR2 of DIAP1 bound to amino-terminal Hid, Reaper, and Grim peptides further indicates a conserved mechanism of binding in insect and mammalian proteins (22). However, a different study indicated that Hid bound to DIAP1 in vitro despite the fact that additional protein sequences, consisting of a glutathione S-transferase tag used for purification, were fused to the amino terminus of Hid, presumably preventing normal amino-terminal processing (18).
Op-IAP has been shown to bind Hid, Reaper, and Grim and inhibit apoptosis stimulated by overexpression of these proteins (9,10). Previous work by Vucic et al. (23) showed that the region of Op-IAP containing BIR2 and its flanking sequences was necessary and sufficient to co-immunoprecipitate Hid and block Hid-induced apoptosis in Sf21 cells, although the ability to inhibit apoptosis was less efficient than full-length Op-IAP. Furthermore, a sequence required for Hid binding was mapped by deletion analysis to a region encompassing residues 174 -190 of Op-IAP, including a portion of the carboxyl-terminal end of BIR2 and some of its downstream flanking sequence. Interestingly, however, this deleted region in BIR2 of Op-IAP does not include the residues that correspond to those in BIR3 of XIAP responsible for interacting with the amino terminus of Smac, as determined by crystallography. In addition, a second deletion mutant lacking residues 157-173 of Op-IAP, which does lack the residues corresponding to those shown in XIAP to bind to Smac, still co-immunoprecipitated with Hid (23). These results, along with those indicating that Hid can bind DIAP1 in vitro without processing at its amino terminus (18), seemed to contradict the hypothesis that Hid binds IAP proteins in a manner similar to that of Smac. Therefore, we decided to investigate the interaction between Hid and Op-IAP more closely.
For bacterial expression of OpIAP BIR2, coding sequence encompassing residues 95-199 of Op-IAP, including an amino-terminal HA tag, was cloned into pET-15b (Novagen). This produced a protein containing a His 6 and HA tag amino-terminal to BIR2 with its flanking sequences.
Bacterial Expression and Purification of Recombinant BIR2-BL21(DE3)pLysS cells transformed with pET-15bBIR2 were used to inoculate 1.5 liters of Luria Broth (LB) containing 100 g/ml ampicillin. This culture was grown to an A 600 of 0.5 at 37°C, and protein expression was induced by the addition of isopropyl-1-thio-␤-D-galactopyranoside to a final concentration of 0.4 mM for 3 h at 25°C. The culture was then centrifuged at 10,000 ϫ g for 10 min at 4°C and the pellet stored overnight at Ϫ80°C to facilitate cell breakage. Once thawed the cells were resuspended in 40 ml of chilled Extraction/Wash Buffer and purified by affinity chromatography using Talon (CLONTECH) metal affinity resin following the batch/gravity-flow protocol provided in the Talon Metal Affinity Resins User Manual (CLONTECH).
Protection Assays-In the actinomycin D protection assay, 6-well plates were seeded at a density of 5 ϫ 10 5 Sf21 cells/well and transfected with 1 g of the indicated plasmid by lipid-mediated transfection. A lacZ-expressing plasmid was used as a negative control. 20 h after transfection cells were heat-shocked at 42°C for 30 min to induce expression. After heat shock, cells were placed back at 27°C. Four hours after the beginning of heat shock, the media were removed and replaced with fresh media containing actinomycin D to a final concentration of 500 ng/ml. The cells were incubated with actinomycin D for 10 h at 27°C and then stained with 1 g/ml Hoechst 33258 dye (Molecular Probes), and all non-apoptotic nuclei were manually scored by fluorescence microscopy. The number of non-apoptotic nuclei were determined and compared with the number of non-apoptotic nuclei of lacZ-transfected controls that were not treated with actinomycin D (set at 100%), and a percent was generated that represented cell viability. This number was then subtracted from 100% and presented as percent apoptosis to be consistent with the remaining figures. Transfection efficiency was monitored in every experiment by transfection of an eGFP-expressing plasmid, and the average transfection efficiency for all experiments was 67%.
In the Hid protection assay, 2 g of the pHidflaghis plasmid was cotransfected with either 2 g of wild type Op-IAP plasmid or the full-length Op-IAP point mutant plasmids. Alternatively, 2 g of the pHidflaghis plasmid was co-transfected with 8 g of the hybrid mutant Op-IAP plasmids to give a 4:1 concentration ratio of mutant Op-IAP to Hid plasmid. Where necessary, an eGFP-expressing plasmid was used to balance the amount of DNA in each transfection. Twenty hours after transfection cells were heat shocked as described above and then incubated at 27°C for 1.25 h. At 1.25 h following the beginning of heat shock, cells exhibiting blebbing were scored as apoptotic. For both types of protection assays, 3 randomly chosen high power fields were counted per well, using 2 wells per condition per assay, and each assay was independently repeated at least 3 times.
Hid Peptide Precipitation Assay-To determine whether bacterially expressed BIR2 could bind to Hid, 10 l of a 1 mg/ml solution of a peptide representing amino acids 2-11 of Hid with a carboxyl-terminal lysine residue and biotin label (AVPFYLPEGGK-biotin) was added to 200 l of a 10% slurry of streptavidin-conjugated agarose beads (Sigma) in Nonidet P-40 lysis buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1.0% Nonidet P-40, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride) and gently rocked for 1 h with the beads at 4°C to allow the peptide to bind to the beads. A similar peptide that included the initiating methionine (MAVPFYLPEGGK-biotin) was used as a control. Peptides were synthesized and purified to Ն95% purity by Sigma. The beads were washed once with 1 ml of Nonidet P-40 lysis buffer to remove any unbound peptide, followed by the addition of 5 g of biotin. 120 ng of purified bacterially expressed BIR2 protein was added to each of the three tubes and incubated with gentle rocking overnight at 4°C. The following day the beads were washed six times with 1 ml of Nonidet P-40 lysis buffer. Protein loading buffer and 1% 2-mercaptoethanol were added, and the mixture was heated to 95°C to elute the complex from the beads. Proteins were analyzed by immunoblotting with anti-HA.11 mouse monoclonal antibody (Babco) and goat anti-mouse IgG horseradish peroxidase-conjugated antibody. Bound antibody was detected using SuperSignal West Pico chemiluminescent reagent (Pierce).
To analyze binding of proteins expressed in Sf21 cells, 6-well tissue culture plates were seeded with 5 ϫ 10 5 cells/well, and 2 wells each were transfected with 5 g of each of the respective plasmids and heatshocked as above. Hid peptide (1 l of a 1 mg/ml solution) was added to streptavidin-conjugated beads and incubated for 1 h at 4°C. The beads and peptide were then washed once with 500 l of Nonidet P-40 lysis buffer, and Nonidet P-40 lysis buffer (600 l) was added to the beads and incubated with rocking at 4°C until use. Three hours following the beginning of heat shock, the cells were harvested in 100 l/well of Nonidet P-40 lysis buffer. The cells of two wells of similar conditions were combined and lysed for 30 min at 4°C. After lysis the cells were sonicated for 10 s and then spun for 5 min at 14,000 rpm at 4°C to remove cellular debris. Forty l of the clarified sample was removed from each sample and transferred to a clean tube, followed by addition of protein loading buffer and 1% 2-mercaptoethanol. The samples were stored at Ϫ20°C until analysis by immunoblotting. The remaining 160 l of clarified supernatant from each tube was added to the beads with the bound Hid peptide and allowed to rock gently at 4°C overnight. The following day the beads were washed three times with 500 l of Nonidet P-40 lysis buffer, and bound proteins were analyzed by immunoblotting as described above. Where indicated, 10 mM EDTA was added to the Nonidet P-40 lysis buffer, and the samples were then processed as described above, or 50 g/ml MG-132 (Calbiochem) was added to the cells 0.5 h prior to heat shock.

Op-IAP Binds Directly to the Processed Amino Terminus of
Hid-It was shown previously that Op-IAP co-immunoprecipitates with overexpressed Hid in Sf21 cells and that this interaction involves the BIR2 domain of Op-IAP and the amino terminus of Hid (10,23). However, direct binding of Op-IAP to Hid has not been demonstrated previously, because co-immunoprecipitation experiments from cell lysates do not exclude the possibility that other proteins are involved in the interaction. Thus, we developed a Hid binding assay that does not rely on proteins expressed in eukaryotic cells. Although full-length Op-IAP proved to be highly insoluble when overexpressed in bacteria, we were able to bacterially express and purify the BIR2 domain of Op-IAP, along with sequences flanking the BIR domain (amino acids 95-199), containing a His 6 sequence and an HA epitope tag at the amino terminus. The purified recombinant BIR2 protein was used in binding assays with two carboxyl-terminally biotinylated peptides consisting of either amino acids 1-11 or 2-11 of the Drosophila Hid protein (Fig.  1A). In this assay, recombinant BIR2 protein bound specifically to the peptide lacking the initiating methionine (amino acids 2-11; hereafter called the Hid peptide), but there was no detectable binding to the peptide containing the initiating methionine (amino acids 1-11; hereafter called the Met-Hid peptide). The results of this assay demonstrate a direct and specific interaction between BIR2 of OpIAP and the processed amino terminus of the Hid protein.
Similar to the bacterially expressed BIR2 protein, full-length Op-IAP protein expressed in Sf21 cells also bound to the Hid peptide (Fig. 1B). Interestingly, however, Op-IAP expressed in Sf21 cells also appeared to bind the Met-Hid peptide, in contrast to the results obtained with bacterially expressed and purified BIR2 protein. To explain these contradictory results, we hypothesized that the methionine residue on the Met-Hid peptide was being removed by methionine aminopeptidase activity in the Sf21 cell lysate during incubation with the Met-Hid peptide. To test this hypothesis we added EDTA to the lysate, which would be expected to inhibit this class of proteases. As predicted, the addition of EDTA to the lysate greatly reduced binding of Op-IAP to the Met-Hid peptide but did not affect binding to the Hid peptide, consistent with the presence of methionine aminopeptidase activity in the lysate (Fig. 1B). Thus Op-IAP expressed in Sf21 cells bound specifically to the processed amino terminus of Hid, as did the BIR2 domain with its flanking sequences expressed in Sf21 cells (Fig. 1C). As expected based on previous results, the BIR1 and RING domains of Op-IAP, along with the negative control chloramphenicol acetyltransferase protein, did not bind to the Hid peptide (Fig. 1C). The Drosophila DIAP1 protein, which has also been reported to co-immunoprecipitate with Hid (10) and bind Hid in vitro (17,18), also bound to the Hid peptide (data not shown).
The Conserved Residues in the BIRs of Op-IAP Are Important for Anti-apoptotic Function-To assess the role of the invariant conserved residues of the BIR region in Hid binding and antiapoptotic function, we determined the effect of mutating a number of these residues to alanine ( Fig. 2A). Furthermore, two deletion mutants were constructed by deleting either BIR1 or BIR2 in the context of the entire Op-IAP protein (Fig. 2B). We then assayed whether expression of these mutated proteins could protect Sf21 cells against different apoptotic stimuli. Expression of the mutated proteins was verified by immuno-blotting (Fig. 3C), whereas transfection efficiency was monitored in each assay using a similar vector expressing eGFP and determining the percentage of fluorescent cells. It should be noted that the level of apoptosis observed upon transfection of wild type Op-IAP was essentially the same as the number of untransfected cells in this assay, suggesting that virtually all of the cells expressing wild type Op-IAP were protected.
Sf21 cells were transiently transfected with constructs expressing the respective point mutants, treated with actinomycin D to induce apoptosis, and apoptotic cell death was quantified (Fig. 3A). All of the conserved residues mutated in both BIRs of Op-IAP were important for the anti-apoptotic function of this protein in the context of actinomycin D-induced death, with the notable exception of the C148A mutation, which for unknown reasons retained partial activity. The construct lacking BIR1 (B1D) also retained partial activity in this assay, whereas deletion of BIR2 (B2D) led to a complete loss of pro-2 M. C. Green and R. J. Clem, manuscript in preparation.
FIG. 1. The processed amino terminus of Hid binds directly and specifically to BIR2 of Op-IAP. A, streptavidin-agarose beads were incubated with biotinylated peptides containing amino acids 1-11 (Met-Hid peptide) or 2-11 (Hid peptide) of Hid or no peptide (Neg. control). Bacterially expressed and purified HA-tagged BIR2 domain of Op-IAP was incubated with the beads, and the protein that specifically bound was eluted and detected by immunoblotting. Input represents 10% of the amount of BIR2 protein added to the beads. Molecular mass markers are shown on the left. B, lysates were prepared from Sf21 cells expressing full-length, HA-tagged Op-IAP protein and were incubated with peptide-bound beads as in A, and bound protein was detected by immunoblotting. In the right-hand panels, EDTA was added to the lysis buffer to inhibit metalloprotease activity. Top panels, protein that bound to the peptides (pull down); bottom panels, immunoblot analysis of the whole cell lysates (WCL). C, lysates from Sf21 cells expressing the indicated HA-tagged proteins were allowed to bind to Hid peptidebound beads or beads alone (Neg. control), and proteins that bound to the Hid peptide were analyzed by immunoblotting. Top and bottom panels represent the results of the pull down and whole cell lysate analysis as described above in B. The RING domain runs as a smear due to ubiquitination. 2 tection, suggesting that BIR2 is more important than BIR1 for protection against actinomycin D-induced apoptosis. However, whereas BIR2 appears to be crucial, the loss of BIR1 also reduced the ability of Op-IAP to protect against actinomycin D, indicating that BIR1 also contributes to protection against this death stimulus.
To test the ability of the mutated proteins to inhibit another death signal that is less potent than actinomycin D, we cotransfected the mutated versions of Op-IAP with a vector expressing full-length Drosophila Hid at a ratio of 1:1 (Fig. 3B). In this assay, a greater difference was observed between point mutations in BIR1 and BIR2 than in the actinomycin D assay. Point mutations in BIR1, such as R21A, G42A, C54A, W64A, and the double mutant G42A/W64A, largely retained the ability to protect against Hid-induced death, whereas point mutations in BIR2, such as R114A, C148A, and H168A, showed significantly less protection, including the double mutation R21A/R114A, which behaved like a single mutation in BIR2 (compare R21A/R114A with R21A and R114A). Consistent with these results, deletion of BIR1 (B1D) had no effect on the ability of Op-IAP to protect against Hid-induced death, whereas deletion of BIR2 (B2D) resulted in decreased protection. However, deletion of BIR2 (B2D) did not result in total loss of anti-apoptotic activity when compared with the vector control, suggesting that BIR1 and/or the RING have some effect on Hid-induced apoptosis. Similar to the actinomycin D results, these results also suggest that BIR2 is more important than BIR1 in the anti-apoptotic activity of Op-IAP and corroborates results published previously (23).
We next correlated the anti-apoptotic activity of the mutated Op-IAP proteins with their ability to bind to the Hid peptide (Fig. 3C). Immunoblot analysis of whole cell lysates showed that all of the mutant proteins were expressed at equivalent levels. All of the proteins containing mutations in invariant residues of BIR1 or BIR2 bound the Hid peptide, with the exception of R21A/R114A and C175A. In addition, reduced binding was observed for the C54A and H168A constructs. The C175A, C54A, and H168A constructs all contain mutations in zinc-coordinating residues in either BIR1 or BIR2, which might be expected to disrupt the folding of these domains. However, the fact that most of the mutant proteins still bound the Hid peptide indicates that they were not completely unfolded or misfolded as a result of the introduced mutations. As expected, B1D, which lacks the entire BIR1 domain, still bound the Hid peptide as well as wild type Op-IAP, whereas B2D (lacking BIR2) did not bind the Hid peptide (Fig. 3C). Residues invariant in all BIRs are highlighted. Residues that were mutated in this study are indicated by the appropriate amino acid below the sequence. A bracket indicates the extent of the deletion in construct ⌬157-173. B, the primary structure of Op-IAP is shown, as is that of deletion constructs lacking BIR1 (B1D) or BIR2 (B2D) and domain swapping constructs with the core sequence of BIR1 surrounded by the amino-and carboxyl-terminal flanking sequences from around BIR2 (BIR1 N2, BIR1C2, and BIR1 NC2) or the core sequence of BIR2 surrounded by the aminoand carboxyl-terminal flanking sequences from around BIR1 (BIR2 N1, BIR2 C1, and BIR2 NC1). Deletions in B1D and B2D are indicated.

The BIR2 Flanking Regions Do Not Confer Anti-apoptotic
Activity to BIR1-Deletions in either of the sequences flanking BIR2 of Op-IAP (deletion of either residues 106 -109 or 174 -190) were shown previously to eliminate the protective function of BIR2 against Hid (23). The 174 -190 deletion also eliminated Hid binding, suggesting that the carboxyl-terminal flanking sequence of BIR2 may be involved in binding Hid (23). Based on these previous results, we wondered whether the sequences flanking BIR2 are what are important for conferring antiapoptotic function to BIR2, and whether the BIR2 flanking sequences could confer anti-apoptotic activity or Hid binding onto another BIR core sequence. To test this, we created hybrid constructs that contained the core sequence of BIR1 surrounded by either one or both of the flanking sequences of BIR2 (BIR1 N2, BIR1 C2, or BIR1 NC2; Fig. 2B), and we then tested whether expression of these hybrid BIRs could protect Sf21 cells against Hid-induced apoptosis at an IAP:Hid ratio of 4:1 (Fig. 4A). Similar to previous results (23), expression of BIR2 and its flanking sequences (OpIAP BIR2; amino acids 95-199) was insufficient to inhibit Hid-induced apoptosis when cotransfected with Hid at a 1:1 ratio (data not shown), but protection was observed when OpIAP BIR2 was co-transfected with Hid at a 4:1 ratio. Also as expected, BIR1 and its flanking sequences (OpIAP BIR1) did not inhibit Hid-induced apoptosis, even at a 4:1 ratio (Fig. 4A). Even though the hybrid constructs containing the flanking regions from BIR2 surrounding BIR1 were expressed at high levels (Fig. 4B), they were unable to inhibit apoptosis that was induced by expression of Hid (Fig.  4A). The BIR1 hybrid proteins also did not bind the Hid peptide (Fig. 4B), indicating that the flanking sequences of BIR2 are not sufficient to confer Hid binding or anti-apoptotic activity onto the BIR1 core sequence.
After having shown that the BIR2 flanking sequences are not sufficient to impart anti-apoptotic activity on a non-functional BIR, the next question we sought to address was whether the BIR2 flanking sequences are indeed specifically required for inhibiting Hid-induced cell death or whether they could be replaced with sequences flanking a non-functional BIR. Therefore, we again created hybrid constructs, except this time the flanking sequences of BIR2 were exchanged for those of BIR1, creating constructs with the core sequence of BIR2 surrounded by the flanking sequences of BIR1 (BIR2 N1, BIR2 C1, and BIR2 NC1; Fig. 2B). When these constructs were co-expressed with Hid at a 4:1 ratio, they were also unable to inhibit Hidinduced apoptosis compared with the BIR2 control (Fig. 5A), indicating that both the core and flanking sequences of BIR2 are necessary for blocking Hid-induced apoptosis. It should be noted that the level of expression of some of the BIR2 hybrid constructs was low (Fig. 5B), which may have affected the results obtained in this experiment. Because Op-IAP is ubiquitinated when expressed in Sf21 cells, 2 we used a proteasome inhibitor to stabilize the levels of the hybrid proteins and then repeated the assay. However, when the expression of the BIR2 flanking mutants was stabilized by the addition of the proteasome inhibitor MG132, they still could not protect against Hid-induced apoptosis even though all three could still bind the Hid peptide ( Fig. 5B and data not shown). It should be noted that MG-132 treatment itself does not induce apoptosis in Sf21 cells (data not shown). The ability of the hybrid proteins to bind the Hid peptide indicates that the folding of the hybrid proteins

FIG. 3. The effect of mutations in either BIR1 or BIR2 of Op-IAP on protection against actinomycin D and Hid-induced apoptosis.
A, constructs expressing the indicated mutated proteins were transfected into Sf21 cells, and the cells were then treated with actinomycin D to induce apoptosis, and the average percentage (Ϯ S.E.) of apoptotic cells was determined. The average transfection efficiency in all experiments was 67%. B, Sf21 cells were transfected with a lacZ-expressing vector (negative control), or vectors expressing the indicated proteins were cotransfected with a Hid-expressing vector at a 1:1 ratio. The average number of apoptotic cells per high power field (Ϯ S.E.) was determined as described under "Experimental Procedures." C, lysates were prepared from Sf21 cells expressing the indicated proteins and analyzed for binding to the Hid peptide (pull down) as described in Fig. 1. was not completely disrupted. In addition, the ability of the BIR2 hybrid constructs to bind the Hid peptide indicates that binding to the processed amino terminus of Hid is solely a property of the core BIR2 domain and does not require specific sequences in the flanking regions. From these results we conclude that both the BIR2 core sequence and the sequences flanking BIR2 are required to block Hid-induced apoptosis, but neither are sufficient for inhibitory activity. Instead, both the core and flanking sequences of BIR2 are required in tandem to inhibit Hid-induced apoptosis. We also conclude that the structural elements required to bind the amino terminus of Hid are confined to the core sequence of BIR2, and that simply binding Hid is not sufficient to block Hid-induced death.
Evolutionary Conservation of Hid-Op-IAP Versus Smac-XIAP Interactions-The processed mammalian protein Smac shares a short region of homology with Hid, Reaper, and Grim at their amino-terminal ends, and the first four residues in Smac (Ala-Val-Pro-Ile) have been shown to contact a groove in BIR3 of XIAP, resulting in a stable interaction (20,21). Five residues lining this groove in BIR3 of XIAP were predicted to be important in creating the chemical interactions with the first four amino acids of processed Smac (21). After comparing BIR2 of Op-IAP to BIR3 of XIAP, we found that these same five residues in BIR3 of XIAP (Gly-306, Leu-307, Trp-310, Glu-314, and Trp-323) were either identical or chemically similar in BIR2 of Op-IAP ( Fig. 2A). Furthermore, by comparing the BIR1 and BIR2 sequences of Op-IAP, we discovered that only one of these five residues was not conserved in BIR1. This residue is Gly-306 in BIR3 of XIAP, Gly-154 in BIR2 of Op-IAP, and Glu-60 in BIR1 of Op-IAP. Therefore, we hypothesized that mutating residue Glu-60 in BIR1 of Op-IAP to a glycine might allow Hid binding to BIR1. However, when this mutation was constructed in OpIAP BIR1 (BIR1 E60G), it did not bind the Hid peptide or inhibit apoptosis (Fig. 6, A and B). We further hypothesized that in order for this E60G change to have an effect on Hid-induced death, it would have to be made in the BIR1 NC2 construct because the presence of the BIR2 flanking sequences may be required for anti-apoptotic activity based on the results in Figs. 4 and 5. However, the BIR1 NC2 E60G construct also could not bind Hid or protect against Hid-induced apoptosis (Fig. 6, A and B). Thus, simply changing the aspartate residue at position 60 to glycine was not enough to confer Hid binding or anti-apoptotic activity to BIR1. Interestingly, however, when the equivalent glycine residue at position 154 in BIR2 was mutated to a glutamate (BIR2 G154E), there was complete loss of anti-apoptotic activity compared with the BIR2 control, accompanied by loss of binding to the Hid peptide (Fig. 6, A and B). Similar results were observed when the same change was made in BIR2 NC1 (data not shown). These results suggest that the amino terminus of Hid binds Op-IAP in a manner very similar to that of Smac binding to XIAP but that there must be other differences between BIR1 and BIR2 in addition to Glu-60 that preclude Hid binding to BIR1.
Given these results, we decided to take a closer look at the  Fig. 2. B, Sf21 cell lysates containing the indicated proteins were analyzed for binding to the Hid peptide as described in Fig. 1. WCL, whole cell lysates; Neg. control, no peptide.  (23) and also eliminated binding to the Hid peptide (Fig. 6B), but it did not include the residues making up the groove that would be predicted to be responsible for contacting the amino-terminal 4 amino acids of Hid. The majority of this deletion is in the carboxyl-terminal flanking sequence of BIR2; however, the most carboxyl-terminal conserved cysteine in BIR2 (Cys-175), which is predicted to be involved in the coordination of a zinc ion, was also removed in this deletion. This prompted us to question whether the deleted region was actually responsible for the interaction with Hid or if the deletion was simply causing a disruption in protein folding by removal of the zinc-coordinating cysteine. To approach this question, Cys-175 was mutated to alanine and then tested in the Hid protection assay for anti-apoptotic activity. As expected based on the lack of protection by other point mutants in BIR2 (Fig. 3B), the C175A mutant did not inhibit Hid-induced apoptosis (Fig. 6A). More interesting, however, was that C175A also lost the ability to bind the Hid peptide (Fig. 3C), suggesting that the removal of this key amino acid in the 174 -190 deletion was the reason for the loss of Hid binding.
Given the importance of the Smac-binding groove in BIR3 of XIAP, we were puzzled by the previous report (23) that deleting residues 157-173 of Op-IAP did not affect Hid binding, because this deletion completely removes the residues making up the predicted Hid-binding groove in BIR2 of Op-IAP. We thus tested the 157-173 deletion in our Hid peptide binding assay, and we found that in this assay removal of these amino acids eliminated binding to the Hid peptide (Fig. 6B). These results, coupled with those of Vuvic et al. (23), suggest that there is a second interaction between HID and BIR2 of Op-IAP that does not involve the amino terminus of HID, which allows the 157-173 deletion to interact with full-length Hid but not with the Hid peptide. This putative second interaction may also explain the binding of glutathione S-transferase-Hid to DIAP1 that was reported previously (18). DISCUSSION In this study, we have shown that a peptide containing the amino terminus of Hid binds directly in vitro to BIR2 of Op-IAP without the need for other proteins but only in the absence of the initiating methionine residue of Hid. We have also shown that either full-length Op-IAP or BIR2 expressed in Sf21 cells binds to this peptide only in the absence of the initiating methionine and that Sf21 cells contain aminopeptidase activity capable of removing the methionine from the Met-Hid peptide.
The core sequence of BIR2 is required for binding to the processed amino terminus of Hid, whereas the sequences flanking the core BIR2 domain are not. However, the sequences flanking BIR2 are required for blocking Hid-induced apoptosis. Interestingly, the mechanism of Hid binding to Op-IAP appears to be very similar to that used by the mammalian proteins Smac and XIAP, involving a residue (Gly-154 in Op-IAP and Gly-306 in XIAP) that lines a groove in XIAP BIR3 that contacts the processed amino terminus of Smac. These results further support the hypothesis that Smac is the functional homolog of Hid in mammalian cells (11,12). This extraordinary conservation of structure and function between insect and mammalian death machinery underscores the importance of this interaction in apoptosis regulation.
We have shown for the first time that the invariant conserved residues in both BIRs of Op-IAP are important for Op-IAP protection against actinomycin D-induced apoptosis, although the level of importance varied somewhat from residue to residue. Mutation of Cys-148 in BIR2 to alanine resulted in only a partial loss of anti-apoptotic function in the actinomycin D assay. This was unexpected because Cys-148 is predicted to be involved in zinc coordination, and we expected that mutation of this residue would have drastic effects on the structural integrity of BIR2. However, the fact that proteins bearing mutations in the zinc-binding residues Cys-148 and His-168 still bound the Hid peptide (Fig. 3C) and still co-immunoprecipitate with full-length Hid (23) leads us to the rather surprising conclusion that mutation of individual zinc-coordinating residues does not necessarily completely disrupt the structure of a BIR. However, mutations in Cys-175 and Gly-154 did eliminate binding to the Hid peptide, indicating that these residues are either important in the interaction between Hid and Op-IAP or that their substitution perturbs the structure of the Hid-binding site in BIR2. Cys-175 is a zinc-coordinating residue, but it is also closer in the predicted structure to the groove predicted to bind the amino terminus of Hid than the other zinc-binding residues in BIR2, and thus mutation of Cys-175 may disturb the folding of the Hid-binding groove more than mutation of the other zinc-coordinating residues in BIR2. Gly-154 is interesting because it is not conserved in BIR1 of Op-IAP, and it is also one of the residues predicted to be a crucial difference between the BIRs of XIAP, because it is not conserved in BIR1 or -2 of XIAP. From their crystallography data, Wu et al. (21) predicted that the residue in XIAP equivalent to Gly-154 (Gly-306) is required to allow tight packing of an isoleucine residue at position 4 of Smac into the groove in BIR3 of XIAP. The equivalent residue in position 4 of Hid is a phenylalanine. Mutation of Gly-154 to FIG. 6. The amino terminus of Hid binds to Op-IAP by a mechanism similar to that of Smac binding to XIAP. A, the indicated constructs were transfected into Sf21 cells at a 4:1 ratio of Op-IAP:Hid, and the average percentage of apoptotic cells (Ϯ S.E.) was determined. B, Sf21 cells transfected with the indicated constructs were lysed, and protein was analyzed for binding to the Hid peptide as described in Fig. 1. WCL, whole cell lysates; Neg. control, no peptide. a glutamate would be predicted to prevent the phenylalanine in position 4 of Hid from fitting into the groove, thus disrupting binding. The fact that the G154E mutation eliminated Hid binding indicates that the contacts between the amino terminus of Hid and BIR2 of Op-IAP are similar to those found between Smac and BIR3 of XIAP.
Another interesting observation that arose from the actinomycin D data was that when the conserved residues in BIR1 were individually mutated, all anti-apoptotic function was lost, but when BIR1 was deleted completely, partial function was retained. This could be contributable to protein misfolding in the point mutants, whereas deletion of the entire BIR1 may allow a more natural fold to occur, and thus partial antiapoptotic function is retained. However, all of the proteins with point mutations in BIR1 (with the exception of the double mutant R21A/R114A, which also has a mutation in BIR2) still bound the Hid peptide, indicating that the mutated proteins were not completely misfolded or unfolded. Whereas deletion of BIR1 reduced protection partially compared with full-length Op-IAP, deletion of BIR2 led to a complete loss of anti-apoptotic function against actinomycin D, indicating that BIR2 plays a key role in protecting against this death stimulus. However, BIR1 also seems to assist in protection against actinomycin D to some degree.
In the context of Hid-induced apoptosis, we found that expression of BIR2, although sufficient to partially inhibit Hidinduced death, did so at a level lower than that of wild type Op-IAP. This result is consistent with previous reports (25,26) that the RING domain of Op-IAP is required for optimal antideath function. The weaker ability of BIR2 to inhibit apoptosis is apparently not due to a problem in protein expression, folding, or stability, because the BIR2 construct expressed high levels of protein that still bound the Hid peptide and co-immunoprecipitated Hid as efficiently as wild type Op-IAP ( Fig. 1B and Ref. 23). The fact that B1D, which contains BIR2 and the RING, protected as well as wild type Op-IAP (Fig. 3B) further suggests that the RING plays an important role in protection against Hid-induced apoptosis.
It does not appear that Op-IAP is functioning by simply binding and sequestering Hid in this system, given the existence of several mutants that still bind Hid but do not block Hid-induced death. However, the ability to bind Hid does appear to be necessary for Op-IAP to block apoptosis induced by overexpression of Hid. This is in contrast to results obtained in Drosophila with DIAP1, in which gain-of-function DIAP1 mutants that have reduced Hid binding are actually better at preventing Hid-induced death than wild type DIAP1 (18). The difference between these two systems may lie partially in the fact that we are overexpressing Hid and Op-IAP. However, it must be remembered that Op-IAP is a viral protein that is not normally present in cells, and so in the context of virus infection, Op-IAP is, by definition, overexpressed. Op-IAP therefore may have a function that is somewhat different from that of cellular IAPs. Both the amino-and carboxyl-terminal sequences flanking BIR2 play a critical role in inhibition of Hidinduced apoptosis. The function of these sequences is not known, but it is apparently not simply to stabilize BIR2 or facilitate proper folding, given the hybrid constructs (containing the core BIR2 sequence and flanking sequences from BIR1) that still bind the Hid peptide but do not inhibit Hid-induced apoptosis. The function of the BIR2 flanking sequences may include caspase binding and inhibition, but a caspase that binds Op-IAP has not yet been identified.
Bacterially expressed, purified BIR2 protein bound only to the Hid peptide lacking the initiation methionine and not to the Met-Hid peptide. However, Op-IAP expressed in Sf21 cells bound to both peptides. The observation that Op-IAP did not bind to the Met-Hid peptide when methionine aminopeptidase activity was inhibited by EDTA confirms that Hid must be processed, presumably by a methionine aminopeptidase, before it can bind to Op-IAP. This is perhaps not surprising, because it is common for the initiating methionine residue to be removed from eukaryotic proteins by methionine aminopeptidases, especially when the second position is occupied by a small amino acid (27), such as in the case of Hid where the second amino acid is alanine. Nonetheless, although it was postulated previously that Hid must be processed to bind IAPs (15), this is the first experimental evidence that the amino terminus of Hid is processed in cells.
It was shown previously (23) that deletion of amino acids 174 -190 in Op-IAP led to loss of Hid binding, suggesting that the flanking region downstream of BIR2 may be involved in the interaction with Hid. Our results with the C175A mutant indicate that the loss of Cys-175 alone in this deletion is sufficient to eliminate binding to the amino terminus of Hid. This combined with the observation that the BIR2 NC1 domain swapping mutant, which contains the core sequence of BIR2 (amino acids 114 -178) but the flanking regions of BIR1, still bound the Hid peptide indicates that the ability to bind the amino terminus of Hid is due to the core BIR2 sequence and does not involve the residues flanking BIR2.
In our assay, deletion of amino acids 157-173 from Op-IAP resulted in a complete loss of binding to the Hid peptide. This is in contrast to the co-immunoprecipitation results of Vucic et al. (23), where this deletion did not affect binding to full-length Hid. The difference in these results could be due to the use of full-length Hid protein in the co-immunoprecipitation experiments versus only amino acids 2-11 in our assay. Thus, it is possible that another domain of the Hid protein in addition to the amino terminus contacts Op-IAP at a second site outside of the 157-173 region and that this second interaction is sufficiently strong to detect by co-immunoprecipitation. In support of this, a second contact site between Smac and BIR3 of XIAP was observed in the co-crystal structure in addition to the amino terminus-binding groove (21). This second contact site involved helices H2 and H3 in Smac and helix ␣1 in BIR3. The equivalent of helix ␣1 in Op-IAP lies at the amino-terminal end of BIR2, outside of the 157-173 deletion, and so a similar second interaction may also exist between Op-IAP and Hid.
In conclusion, we have shown for the first time that the interaction between Drosophila Hid and baculovirus Op-IAP is highly similar to the interaction between human Smac and XIAP. This interaction is required for the ability of Op-IAP to inhibit Hid-induced apoptosis, but the interaction alone is not sufficient to block death. A complex set of interactions between Op-IAP, Hid, and presumably other as yet unidentified proteins appears to be involved in the anti-apoptotic function of Op-IAP. These unidentified proteins may include caspases, E2 ubiquitin-conjugating enzyme factors, and/or other proteins.