SMAC Negatively Regulates the Anti-apoptotic Activity of Melanoma Inhibitor of Apoptosis (ML-IAP)*

Inhibitors of apoptosis (IAPs) physically interact with a variety of pro-apoptotic proteins and inhibit apoptosis induced by diverse stimuli. X-linked IAP (X-IAP) is a prototype IAP family member that inhibits several caspases, the effector proteases of apoptosis. The inhibitory activity of X-IAP is regulated by SMAC, a protein that is processed to its active form upon receipt of a death stimulus. Cleaved SMAC binds X-IAP and antagonizes its anti-apoptotic activity. Here we show that melanoma IAP (ML-IAP), a potent anti-cell death protein and caspase inhibitor, physically interacts with SMAC through its BIR (baculovirus IAP repeat) domain. In addition to binding full-length SMAC, ML-IAP BIR associates with SMAC peptides that are derived from the amino terminus of active, processed SMAC. This high affinity interaction is very specific and can be completely abolished by single amino acid mutations either in the amino terminus of active SMAC or in the BIR domain of ML-IAP. In cells expressing ML-IAP and X-IAP, SMAC coexpression or addition of SMAC peptides abrogates the ability of the IAPs to inhibit cell death. These results demonstrate the feasibility of using SMAC peptides as a way to sensitize IAP-expressing cells to pro-apoptotic stimuli such as chemotherapeutic agents.

Programmed cell death, or apoptosis, is a genetically regulated mechanism that plays an important role in development and homeostasis in metazoans (1). Abnormalities in programmed cell death that lead to early cell death or the absence of normal cell death have been linked to a variety of human diseases, including neurodegenerative disorders and cancer (2). Currently there are two well characterized apoptotic pathways, one initiated through the engagement of cell surface death receptors by their specific ligands (3) and the other triggered by changes in internal cellular integrity (4). Both pathways eventually converge, resulting in activation of caspases, cysteinedependent aspartate-specific proteases that comprise the effector arm of the apoptotic process (5).
The major regulators of caspases are the IAPs 1 or inhibitors of apoptosis (6). Originally identified in baculoviruses by their ability to substitute functionally for P35, a potent anti-apo-ptotic gene product (7)(8)(9), IAPs have been discovered in both invertebrates and vertebrates (10 -18). Members of the IAP family are characterized by one to three tandem baculovirus IAP repeat (BIR) motifs, and most of them also possess a carboxyl-terminal RING finger motif (6). IAPs inhibit apoptosis induced by a variety of stimuli and interact with multiple cellular partners (19). The anti-apoptotic activity of several IAPs has been attributed to their ability to inhibit caspases (15,20,21). Human X-chromosome-linked IAP (X-IAP), for example, inhibits active caspases-3 and -7 and Apaf-1-cytochrome c-mediated activation of caspase-9 (22,23). This inhibitory activity is mediated through distinct BIR domains of X-IAP; the BIR2 domain and preceding linker region inhibit caspases-3 and -7, while BIR3 blocks caspase-9 (24 -26). Similarly, the anti-apoptotic activity of melanoma IAP, or ML-IAP, is attributed to its lone BIR domain (15). ML-IAP can bind and inhibit caspase-9 through its BIR domain, and mutations in the BIR reduce both inhibition of caspase-9 and general anti-apoptotic activity (15).
IAPs are themselves regulated by proteins that block their anti-apoptotic activity (27). In Drosophila, Reaper (RPR), HID, and GRIM physically interact with and inhibit the anti-cell death activity of D-IAP1 and D-IAP2, fly members of the IAP family (28 -30). SMAC/DIABLO performs a similar function to RPR, HID, and GRIM in mammals (31,32). An amino-terminal signal sequence targets SMAC to mitochondria (31), but during apoptosis, SMAC is processed into the active form and released into the cytosol where it binds IAPs and prevents them from inhibiting caspases (31,33). Thus, SMAC, by binding to the BIR2 and BIR3 domains of X-IAP, abrogates inhibition of the caspases-3 and -9 (34). Interestingly, the only sequence homology between insect RPR, HID, and GRIM and human SMAC is in the four amino-terminal residues of the active proteins (35). The same region is present in the linker peptide of processed caspase-9 (35) and in HtrA2, a recently identified SMAC-like IAP antagonist (36 -39). This short peptide fits into a small hydrophobic pocket on the surface of the X-IAP BIR3 domain and is essential for binding IAPs and blocking their caspase inhibitory activity (34,40,41).
In this study we demonstrate that SMAC physically interacts with ML-IAP and abrogates the ability of ML-IAP to inhibit apoptosis. Peptides derived from the amino terminus of active SMAC are also shown to bind ML-IAP and attenuate its anti-apoptotic activity. The specificity of the SMAC-ML-IAP interaction is supported by the finding that single amino acid changes in the amino terminus of active SMAC or the BIR domain of ML-IAP completely abolish their association.

EXPERIMENTAL PROCEDURES
Expression Constructs-Plasmids expressing ␤-galactosidase, p35, Myc-XIAP, and Myc-ML-IAP, as well as ML-IAP deletions and site-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Fluorescence polarization experiments were performed in 96-well plates on the Analyst HT 96 -384 (Molecular Devices Corporation).
Binding experiments were performed using 1:3 serial dilutions of ML-IAP BIR and X-IAP BIR3 domains starting from 300 M in 50 mM Tris buffer (pH 7.2), 120 mM NaCl, 5 mM dithiothreitol. An approximately 1 nM concentraton of each 5-carboxyfluorescein-tagged probe was added to a set of wells containing the protein dilutions. The K d values of the probe-protein interactions were calculated using Klotz plots and confirmed by Scatchard analysis.
Three-dimensional Modeling and Sequence Analysis-The high resolution crystal structure of SMAC/DIABLO complexed with the BIR3 domain of X-IAP (40) was used to model the NH 2 -terminal four residues (Ala-Val-Pro-Ile) of SMAC/DIABLO into the binding groove of the previously modeled structure of ML-IAP (15). Both protein structures, X-IAP and ML-IAP, were superimposed by their ␣ carbons (root mean square of C␣ ϭ 0.47 Å), and docking of the SMAC peptide was manually performed on the ML-IAP binding site. The ML-IAP⅐SMAC complex was energy minimized using DISCOVER (Molecular Simulations, Inc.). Amino acid sequence alignments were performed using ClustalW (44).

ML-IAP Physically Interacts with SMAC-SMAC has been
shown to physically associate with several IAP family members, most prominently with X-IAP (31). To determine whether SMAC can bind ML-IAP, we coexpressed SMAC with X-IAP, ML-IAP, or vector control. Upon overexpression, SMAC was processed to its active form whereby the first amino-terminal 55 amino acids are cleaved (Fig. 1A). Association of ML-IAP and X-IAP with active SMAC was demonstrated by immunoprecipitation, and the interactions occurred with similar efficiency (Fig. 1A). We investigated the portion of ML-IAP that is responsible for interaction with SMAC using truncation mutants of ML-IAP containing either the BIR or the RING finger domains (Fig. 1B). The BIR domain, like full-length ML-IAP, immunoprecipitated active SMAC but the RING finger domain did not (Fig. 1B). Thus, ML-IAP physically interacts with SMAC through its BIR domain.
ML-IAP BIR Binds RPR-like Peptides-Sequence comparison of the active forms of human SMAC, HtrA2, and caspases-9 and fly RPR, HID, and GRIM revealed a strong similarity in their amino termini ( Fig. 2A) (35). Since the four amino-terminal residues (Ala, Val, Pro, Ile) of active SMAC fit into the binding groove on the surface of the X-IAP BIR3 (40), we investigated whether a similar complex might form between ML-IAP BIR and SMAC peptide. A three-dimensional model based on the reported structure of the SMAC⅐X-IAP BIR3 complex predicted that SMAC peptide should bind ML-IAP BIR much as it does X-IAP BIR3 (Fig. 2B).
To test the validity of our model, we examined whether SMAC, HtrA2, caspase-9, or control peptides could bind the BIR3 domain of X-IAP or the BIR of ML-IAP. SMAC, HtrA2, and caspase-9 peptides efficiently pulled down purified X-IAP BIR3 and ML-IAP BIR, whereas a control peptide did not (Fig.  3A). The binding affinities of these peptides for X-IAP BIR3 and ML-IAP BIR were determined by a fluorescence polarizationbased assay (Fig. 3, B and C). SMAC, HtrA2, and caspase-9 peptides exhibited binding affinities in the low micromolar range for X-IAP BIR3 (Fig. 3B) and in the submicromolar range for ML-IAP BIR (Fig. 3C). These results indicate that ML-IAP BIR binds SMAC and other mammalian RPR-like peptides with high affinity and in a manner similar to X-IAP BIR3.
SMAC Blocks the Anti-apoptotic Activity of ML-IAP-Processing of SMAC exposes the four amino-terminal residues that mediate binding to X-IAP and are required for SMAC to block caspase inhibition by X-IAP. We investigated whether this region of SMAC is also required for SMAC to bind ML-IAP and abrogate its anti-cell death activity. Active SMAC was mimicked in coimmunoprecipitation experiments using an amino-terminally truncated form of SMAC (amino acids 56 -239) (SMAC55M) (Fig. 4A). Full-length SMAC that was processed to its active form was able to bind ML-IAP or X-IAP (Fig. 4B). In contrast, there was no interaction between SMAC55M and ML-IAP or X-IAP (Fig. 4B). Consistent with these results, SMAC but not SMAC55M was able to abrogate ML-IAP-mediated inhibition of adriamycin-induced apoptosis (Fig. 4C). Addition of a single methionine to the amino terminus of active SMAC was therefore sufficient to block its inhibitory effect on ML-IAP.
We also tested whether peptides corresponding to the nine  Fig. 1A. C, MCF7 cells were transiently transfected with the reporter plasmid pCMV-␤gal and either vector control alone or ML-IAP plus vector, SMAC, or SMAC55M. Following transfection, cells were exposed to adriamycin, stained with 5-bromo-4-chloro-3-indolyl ␤-D-galactopyranoside, and apoptosis assessed as described previously (42). D, MCF7 cells were transiently transfected with the reporter plasmid pCMV-␤gal and either vector control, X-IAP, or ML-IAP. Following transfection, SMAC or M-SMAC peptides (50 M) were added where indicated and cells exposed to adriamycin. Apoptosis was assessed as described previously (42). residues at the amino terminus of active SMAC would reverse ML-IAP-mediated inhibition of apoptosis. SMAC and M-SMAC peptides, the latter having an additional amino-terminal methionine, were synthesized as fusions with antennapedia peptide. Antennapedia peptides permit chimeric fusions to gain entry into the cell where they can engage their targets (45). Expression of ML-IAP or X-IAP efficiently blocked adriamycininduced apoptosis, and addition of M-SMAC peptides did not have a significant inhibitory effect on the protective activity of IAPs (Fig. 4D). However, addition of SMAC peptides almost completely negated the ability of ML-IAP and X-IAP to inhibit apoptosis (Fig. 4D). Thus, coexpression of full-length SMAC or addition of SMAC-like peptides abrogates the anti-apoptotic activity of ML-IAP.
SMAC Disrupts Binding of ML-IAP to Processed Caspase-9 -To better understand the mechanism by which SMAC antagonizes the anti-apoptotic function of ML-IAP, we investigated the effect of SMAC on the ability of ML-IAP to bind caspase-9. When overexpressed, caspase-9 undergoes autocatalytic processing, and it is the processed form that physically interacts with X-IAP (35,46). Similarly, ML-IAP coimmunoprecipitated processed caspase-9 but not its zymogen precursor (Fig. 5A). The interaction between ML-IAP and caspase-9 is highly specific, because mutation of aspartate 138 to alanine in the BIR domain of ML-IAP completely abolished the ability of ML-IAP to bind processed caspase-9 (Fig. 5A).
Mutation in the Binding Pocket of ML-IAP Abrogates Interaction with SMAC-Previously, we demonstrated that mutating aspartates 120 and 138 to alanine in ML-IAP abolishes its anti-apoptotic activity (15). The effect of these mutations on ML-IAP interaction with SMAC was characterized. As an additional control we expressed a double glutamate ML-IAP mutant (E87A,E88A) that possessed equivalent anti-apoptotic activity to wild type ML-IAP. SMAC immunoprecipitated ML-IAP and E87A,E88A ML-IAP, but no interaction was observed between SMAC and the D120A or D138A mutants (Fig. 6A). The same was true in the inverse experiment where ML-IAP and E87A,E88A mutant, but not D138A mutant, immunoprecipitated SMAC (Fig. 6B).
To examine the interaction of endogenous SMAC with ML-IAP, we generated stably transfected MCF-7 cell lines expressing FLAG-tagged ML-IAP or the D138A mutant. Consistent with our earlier results (Fig. 6, A and B), endogenous SMAC was coimmunoprecipitated from cells expressing ML-IAP, but not the D138A mutant (Fig. 6C). To determine whether SMAC peptide can bind IAPs expressed in cells, lysates prepared from 293T cells transfected with the X-IAP, ML-IAP, or ML-IAP D138A mutant were incubated with biotinylated SMAC peptide or control peptide. Immunoblotting following peptide precipitation revealed that the SMAC peptide precipitated X-IAP and ML-IAP but not the ML-IAP D138A mutant (Fig. 6D). Therefore, aspartate 138 in the BIR domain of ML-IAP is a critical residue for the binding of SMAC.
Inhibition of caspases by IAPs occurs at the core of the apoptotic machinery, and thus regulation of IAPs by SMAC and SMAC-like proteins represents a key control point in deciding cell fate. We have shown that ML-IAP is regulated by SMAC, since SMAC physically associates with ML-IAP and abrogates the anti-apoptotic activity of ML-IAP. Interaction with SMAC is mediated through the BIR domain of ML-IAP and aminoterminal residues of active SMAC. Three-dimensional modeling together with protein binding studies demonstrated that SMAC binds the BIR of ML-IAP with high affinity and in a manner similar to which it binds the X-IAP BIR3 domain. Further highlighting the similarity of these interactions, mutation of residues in ML-IAP BIR that correspond to functionally important amino acids in X-IAP BIR3 (41) interrupted binding to SMAC. Aspartate 138 of ML-IAP is predicted to be in contact with alanine at the amino terminus of active SMAC peptide and, therefore, is critical for the interaction of ML-IAP with SMAC.
Previous studies have shown the importance of the amino terminus of active SMAC for binding X-IAP (33,40,41). We have demonstrated that this short region is important for binding ML-IAP and antagonizing its anti-apoptotic activity. We have also demonstrated that SMAC peptides can specifically bind and inhibit IAPs. SMAC peptide bound purified BIR domains and X-IAP and ML-IAP from cell lysates, and in doing so, nullified IAP's ability to inhibit apoptosis. To our knowledge, this is the first report in which the functional potential of SMAC or SMAC-like peptides has been explored in a cellular context. Such peptides have obvious therapeutic potential in the treatment of cancer cells that resist conventional cytotoxic therapies. IAPs do contribute to the resistance of cancers to chemotherapeutic agents, since they are widely and, in some cases like ML-IAP, specifically expressed in human malignancies (15,19).