A single BIR domain of XIAP sufficient for inhibiting caspases.

The inhibitor of apoptosis proteins (IAPs) constitute an evolutionarily conserved family of homologous proteins that suppress apoptosis induced by multiple stimuli. Some IAP family proteins, including XIAP, cIAP-1, and cIAP-2, can bind and directly inhibit selected caspases, a group of intracellular cell death proteases. These caspase-inhibiting IAP family proteins all contain three tandem BIR domains followed by a RING zinc finger domain. To determine the structural basis for caspase inhibition by XIAP, we analyzed the effects of various fragments of this IAP family protein on caspase activity in vitro and on apoptosis suppression in intact cells. The RING domain of XIAP failed to inhibit the activity of recombinant caspases-3 or -7, whereas a fragment of XIAP encompassing the three tandem BIR domains potently inhibited these caspases in vitro and blocked Fas (CD95)-induced apoptosis when expressed in cells. Further dissection of the XIAP protein demonstrated that only the second of the three BIR domains (BIR2) was capable of binding and inhibiting these caspases. The apparent inhibition constants (Ki) for BIR2-mediated inhibition of caspases-3 and -7 were 2-5 nM, indicating that this single BIR domain possesses potent anti-caspase activity. Expression of the BIR2 domain in cells also partially suppressed Fas-induced apoptosis and blocked cytochrome c-induced processing of caspase-9 in cytosolic extracts, whereas BIR1 and BIR3 did not. These findings identify BIR2 as the minimal caspase-inhibitory domain of XIAP and indicate that a single BIR domain can be sufficient for binding and inhibiting caspases.


IAP 1 family proteins have been identified in baculoviruses,
Drosophila, and humans, where they function as suppressors of apoptosis (1)(2)(3)(4)(5)(6)(7)(8). In all cases examined thus far, apoptosis has been shown to depend upon the activation of a group of intracellular cysteine proteases with specificity for aspartic acid in the P 1 position of substrates, called caspases (reviewed in Ref. 20). The human XIAP, cIAP-1, and cIAP-2 proteins can bind to and directly inhibit selected caspases. The caspases identified as targets of these IAP family proteins thus far include caspases-3, -7, and -9 but not caspases-1, 2, 8, or 10 (9, 10, 24). 2 XIAP, cIAP-1, and cIAP-2 all contain three tandem copies of a ϳ70-amino acid motif termed the BIR (baculovirus IAP repeat) domain, reflecting the historical discovery of IAP family proteins originally in baculoviruses by Miller and colleagues (1,11). These IAP family proteins also contain a RING zinc finger domain near their C termini. The IAP family protein NAIP, in contrast, contains three tandem BIRs but lacks the RING domain (3). An additional member of the human IAP family, survivin, contains only a single BIR domain but nevertheless can suppress apoptosis in cells (8). The IAP family proteins of Drosophila and the insect-infecting baculoviruses all contain two or three tandem copies of the BIR domain followed by a RING domain (1,2,11). Deletional analyses of the Drosophila and baculovirus IAP family proteins have produced conflicting results, with some studies suggesting a critical role for both the BIR and RING domains and others implying that BIRs are directly involved in apoptosis suppression with the RING motif functioning possibly as a negative regulatory domain (2,12).
The structural similarities and differences among IAP family proteins prompted us to address the question of what constitutes the minimal domain necessary for inhibition of caspases and suppression of apoptosis. Using the human XIAP protein as a model, we determined that only one of the three BIR domains present within this protein was clearly capable of inhibiting caspases-3 and -7, as well as suppressing apoptosis when expressed in cells. These observations demonstrated that a single BIR domain can be sufficient for binding and inhibiting caspases and imply that all BIR domains may not be functionally equivalent, despite their extensive amino acid sequence identity.
Production of GST Fusion Proteins-For production of BIR1ϩ2ϩ3 and RING, pGEX4T-1-XIAP mutant plasmids were introduced into Escherichia coli strain BL21(DE3) containing the plasmid pT-Trx (13). BIR1, BIR2, and BIR3 were produced in XL-1 blue cells (Stratagene, Inc). The GST fusion proteins were prepared from the soluble fraction upon induction with 0.4 mM isopropyl-1-thio-␤-D-galactopyranoside at 30°C for 12 h, affinity purified using glutathione-Sepharose, and then dialyzed against phosphate-buffered saline. Coomassie staining analysis of GST fusion proteins following SDS-PAGE revealed Ն80% intact protein in all cases except BIR3 (see text). His 6 -tagged caspase-3 and -7 recombinant proteins are prepared as described (14,15).
In Vitro Caspase Binding Assays-2 g of His 6 -tagged caspases-3 and -7 were immobilized on 4 l of Ni 2ϩ -Sepharose and incubated with 5 l of in vitro translated, [ 35 S]L-methionine-labeled BIR1, BIR2, BIR2ϩ3, BIR1ϩ2, or BIR1ϩ3 proteins that were prepared from pcDNA3-myc (XIAP mutants) templates using a coupled transcription/ translation system involving rabbit reticulocyte lysates (TNT-lysate, Promega, Inc.) and T7 RNA polymerase. Incubations were performed in 0.1 ml of caspase buffer, consisting of 50 mM Tris-Cl (pH 7.5), 100 mM KCl, 10% sucrose, 0.1% CHAPS, and 5 mM dithiothreitol at 4°C for 1 h. The beads were washed with 1 ml of caspase buffer three times before resuspending samples in boiling Laemmli buffer and analyzing by SDS-PAGE using 750 mM Tris/15% polyacrylamide gels and autoradiography.
Cytochrome c Activation of Caspases in Cytosolic Extracts-Cytosolic extracts were prepared using 293 cells as described (10). 35 S-Labeled in vitro translated pro-caspase-9 (1 l) was added to 5 l of extracts, and caspase activation was initiated by addition of 10 M horse heart cytochrome c (Sigma) and 1 mM dATP at 30°C.
Cell Transfections and Apoptosis Assays-Subconfluent 293 cells were transfected in 3-cm-diameter wells using a calcium phosphate method with 0.5 g of the green fluorescence protein (GFP) marker plasmid pEGFP (CLONTECH), 1 g of pCMV-Fas expression plasmid (16), and 5 g of either pcDNA3-myc-XIAP mutant plasmids or the control plasmid pcDNA3-myc tag. After culturing ϳ1 day, both floating and attached cells were collected, fixed in 3.7% formaldehyde in phosphate-buffered saline for 10 min, and stained with 10 g/ml of 4Ј,6diamidino-2-phenylindole (DAPI). The percentage of apoptotic cells revealed by DAPI staining was counted among the GFP-positive cells by fluorescence microscopy.
Immunoblot Assays-Transfected 293 cells were lysed in 10 mM HEPES (pH 7.5), 142 mM KCl, 1 mM EGTA, 1 mM dithiothreitol, 0.2% Nonidet P-40, 0.1 mM phenylmethylsulfonyl fluoride and used for immunoblot analysis (9,10). Samples were normalized for total protein content (25 g), and immunoblot analysis was performed using 750 mM Tris/12-15% polyacrylamide gels and anti-Myc tag monoclonal antibody 9E10 (Santa Cruz, Inc.) with emission chemiluminescence-based detection (9, 10). Fig. 1 depicts the fragments of XIAP that were expressed as GST fusion proteins in bacteria and affinity purified. The roles of the BIR and RING domains in inhibition of caspases were examined using a C-terminal truncation mutant of XIAP that contained the three tandem BIRs but lacked the downstream linker region and RING domain (BIR1ϩ2ϩ3) and a XIAP fragment lacking the N-terminal BIR domain but retaining the linker and RING regions (RING). Comparison were made of the effects of the BIR1ϩ2ϩ3 (residues 1-336) and RING (residues 337-497) fragments on recombinant caspase-3-or caspase-7mediated cleavage of a fluorigenic tetrapeptide substrate, DEVD-AFC, in vitro. As shown in Fig. 2 (A and B), the BIR1ϩ2ϩ3 fragment of XIAP inhibited DEVD-AFC substrate cleavage by caspases-3 and -7, whereas the C-terminal fragment of XIAP containing the RING domain did not.

RESULTS AND DISCUSSION
Next, fragments of XIAP containing each of the individual BIR domains (Fig. 1) were expressed in bacteria, purified, and tested for inhibition of recombinant caspases-3 and -7 in vitro. Of the three recombinant BIR domains, only BIR2 prevented DEVD-AFC substrate hydrolysis by caspases-3 and -7 (Fig. 2, C  and D). However, because less than half of the purified BIR3 domain appeared to be intact, an additional BIR3-containing fragment of XIAP, which proved to be more stable and which combined BIR1 and BIR3 (Fig. 1), was also produced and tested. This BIR1ϩ3 fragment of XIAP was ineffective at inhibiting caspases-3 or -7 in vitro, whereas BIR1ϩ2 and BIR2ϩ3 fragments that contained BIR2 completely suppressed hydrolysis of DEVD-AFC under the conditions of these in vitro protease assays (not shown).
Determination of the apparent K i for inhibition of recombinant caspases-3 and -7 by the BIR2 fragment of XIAP (residues 124 -260) produced values of 1-5 nM, thus demonstrating potent suppression of these proteases (Table I). Similarly, the BIR1ϩ2 and BIR2ϩ3 fragments of XIAP that contained the BIR2 domain also inhibited caspases-3 and -7 with K i values of 1-3 nM. Thus, although not as potent as the full-length XIAP molecule that has been reported to inhibit these proteases with K i values of 0.2-0.7 nM (10), the BIR2 domain nevertheless retains substantial anti-caspase activity. By comparison, the commonly used tetrapeptide inhibitor DEVD-aldehyde has been reported to inhibit caspases-3 and -7 with K i values of 0.5-3.5 nM (17)(18)(19).
In addition to exploring their ability to inhibit the activity of caspases, the binding of recombinant caspases-3 and -7 to BIR2 and other fragments of XIAP was examined using His 6 -tagged caspases immobilized on Ni 2ϩ resin and 35 S-labeled in vitro translated XIAP fragments. All BIR2-containing fragments of XIAP, including BIR2, BIR1ϩ2, and BIR2ϩ3, bound to immobilized caspases-3 and -7 (Fig. 3A). In contrast, fragments of XIAP lacking BIR2, including BIR1 and BIR1ϩ3, failed to bind these caspases. Note that under these in vitro conditions where high concentrations of proteases are involved, caspase-3 cleaved the BIR2-containing fragments of XIAP, causing slightly faster migration of the 35 S-labeled XIAP protein fragments in gels, whereas caspase-7 did not cleave these same XIAP fragments (Fig. 3A) A Single BIR Domain of XIAP Inhibits Caspases 7788 been obtained to date indicating that XIAP cleavage by caspases occurs under physiological circumstances in intact cells. The specificity of the in vitro binding results was confirmed by experiments in which other His 6 -tagged, recombinant proteins, including caspase-6, caspase-8, and TRAF-3 were tested and found not to bind the BIR2-containing fragment of XIAP (Fig. 3B).
Recently, it has been demonstrated that release of cytochrome c from mitochondria occurs during apoptosis and results in caspase activation by inducing formation of a protein complex containing Apaf-1 and pro-caspase-9 (reviewed in Ref. 21). Moreover, we have found that XIAP, cIAP-1, and cIAP-2 can bind and inhibit pro-caspase-9, preventing it from becoming processed when exposed to cytochrome c and Apaf-1 (24). We therefore tested whether BIR2 was sufficient for preventing cytochrome c-induced processing of pro-caspase-9 in cytosolic extracts. As shown in Fig. 4, full-length XIAP and the BIR2-containing fragment of XIAP completely blocked cytochrome c-induced processing of 35 S-labeled in vitro translated procaspase-9 in cytosolic extracts, whereas BIR1, BIR3, and RING-containing fragments of XIAP were ineffective. Thus, the BIR2 domain of XIAP appears to be sufficient for blocking the   (nM)) for GST-XIAP fragments measured against caspases-3 and -7 Each purified GST fusion protein was incubated at concentrations ranging from 0.7 to 20 nM with either caspase-3 (50 pM) or caspase-7 (300 pM). Progression curves of caspase activity were then measured at 37°C for 30 min using 100 M DEVD-AFC as a substrate. K i app was calculated as previously described (10). Neither BIR1 nor BIR3 showed significant influence on caspase-3 or caspase-7 activity even at 10 M; thus inhibition constants for these polypeptides were estimated to be greater than 1 M.
In vitro binding of BIR2 domain of caspases-3 and -7. His 6 -tagged caspases-3 or -7 (2 g) were immobilized on Ni 2ϩ resin and incubated with in vitro translated 35 S-labeled Myc-tagged XIAP fragments as indicated. After extensive washing, the bound proteins were eluted into boiling Laemmli buffer and analyzed by SDS-PAGE. As a control, an equivalent amount of input 35 S-XIAP protein fragments was loaded directly into gels. In A, cleavage of BIR2-containing fragments by caspase-3 reduced the amounts of XIAP-protein fragments recovered and accounts for the slightly faster mobility of these fragments in gels, compared with input (control) XIAP fragments. In B, the BIR2 fragment was tested for binding to His 6 -caspases-6, -7, and -8 as well as His 6 -TRAF3, demonstrating specific binding to caspase-7.
A Single BIR Domain of XIAP Inhibits Caspases 7789 earliest step in the apoptotic pathway triggered by cytochrome c, i.e. processing of pro-caspase-9. The effects of the BIR2 portion of XIAP on apoptosis in cells were examined by expressing this or other regions of XIAP in 293 cells, which were induced to undergo apoptosis by overexpression of Fas. Transient transfection of 293 cells typically resulted in over half the cells becoming apoptotic, as determined by DAPI staining and UV microscopic analysis (Fig. 5A). In contrast, co-expressing Fas with plasmids producing Myctagged full-length XIAP, BIR1ϩ2ϩ3, or BIR2 alone significantly reduced the percentage of transfected cells undergoing Fas-induced apoptosis. Although the BIR2 fragment was less potent at suppressing Fas-induced apoptosis than full-length XIAP or the BIR1ϩ2ϩ3 fragment of XIAP, it also may have been produced at lower levels in 293 cells (Fig. 5B). In contrast, co-expression of Fas with plasmids producing Myc-tagged fragments of XIAP that lacked BIR2, including myc-BIR1 and myc-BIR1ϩ3, failed to suppress apoptosis (not shown). These BIR2-lacking XIAP mutants, however, were produced at levels comparable with or greater than those of myc-BIR2, based on control transfections experiments using 293 cells that were not induced to undergo apoptosis by Fas (data not presented). Taken together, these findings support the idea that the BIR2 region of XIAP is necessary and sufficient for inhibiting caspases and cellular apoptosis.
Although most IAP family proteins contain two or three copies of the BIR domain, the data provided here suggest that not all BIRs are equivalent in their ability to bind and inhibit caspases. In XIAP, it appears that only the second BIR domain is capable of inhibiting active caspases-3 and -7. Despite their extensive amino acid sequence similarity to BIR2, the BIR1 and BIR3 domains did not inhibit these caspases. Although the issue of protein folding when dealing with mutants is always a potential concern, these findings nevertheless suggest that different BIR domains within IAP family protein may play unique roles in the suppression of cell death. An intriguing idea is that different BIRs may target different subsets of caspases, although no direct evidence for this is yet available. Alternatively, it has been suggested that BIR domains represent a novel type of protein interaction domain. Miller and colleague have noted interactions of the baculovirus IAP family proteins with the Drosophila cell death proteins reaper and doom (22,23), neither of which are caspases. Efforts to identify the targets of each of the individual BIR domains within IAP family proteins and to understand the structural basis for their differential inhibition of caspases will provide additional insights into the mechanisms by which IAPs suppress apoptosis.