Distinct BIR Domains of cIAP1 Mediate Binding to and Ubiquitination ...

Distinct BIR Domains of cIAP1 Mediate Binding to andUbiquitination of Tumor Necrosis Factor Receptor-associatedFactor 2 and Second Mitochondrial Activator of Caspases*□S

Received for publication, August 25, 2005, and in revised form, November 10, 2005 Published, JBC Papers in Press, November 10, 2005, DOI 10.1074/jbc.M509381200

Temesgen Samuel1, Kate Welsh, Thomas Lober, Summanuna H. Togo, Juan M. Zapata, and John C. Reed2

From the Burnham Institute for Medical Research, La Jolla, California 92037

Inhibitor of apoptosis proteins (IAPs) regulate apoptosis primar-ily by inhibiting caspase-family proteases.However,many IAPs alsopossess E3 ligase (ubiquitin-protein isopeptide ligase) activitiesimplicated in both caspase-dependent and -independent functionsof these proteins. Here, we compared the structural features ofcIAP1 responsible for its interactions with two known target pro-teins, TRAF2 and SMAC. The N-terminal (BIR1) and C-terminal(BIR3) BIR domains of cIAP1 were determined to be necessary andsufficient for binding TRAF2 and SMAC, respectively. Mutationalanalysis of the BIR1 and BIR3 domains identified critical residuesrequired for TRAF2 and SMAC binding. Using these mutants,cIAP1-mediated ubiquitination of TRAF2 and SMAC in vitro wasdetermined to be correspondingly dependent on intact bindingsites onBIR1andBIR3.BecauseTRAF2 regulatesNF-�Bactivation,the effects of cIAP1 on TRAF2-mediated induction of NF-�B tran-scriptional activity were studied using reporter gene assays. Expres-sion of a fragment of cIAP1 encompassing the three BIR domains(but not full-length cIAP1) greatly enhanced TRAF2-inducedincreases in NF-�B activity, providing a convenient assay for mon-itoring BIR-dependent effects of cIAP1 on TRAF2 in cells. BIR1mutants of the BIR1–3 fragment of cIAP1 that failed to bindTRAF2lost the ability to modulate NF-�B activity, demonstrating arequirement for BIR1-mediated interactions with TRAF2. Alto-gether, these findings demonstrate the modularity and diversifica-tion of BIR domains, showing that a single cIAP can direct its E3ligase activity toward different substrates and can alter the cellularfunctions of different protein targets in accordancewith differencesin the specificity of individual BIR domains.

Inhibitor of apoptosis proteins (IAPs)3 are a family of anti-apoptoticproteins characterized by the presence of baculoviral IAP repeat (BIR)domains (for view, see Ref. 1). IAPs modulate apoptosis by binding andinactivating caspases (2). Several IAPs contain E2 binding RINGdomains and also operate as E3 ligases that catalyze attachment of ubiq-

uitin to protein substrates. This E3 ligase activity can be directed eithertoward caspases and caspase regulators or toward signal-transducingproteins primarily involved in activation of NF-�B and Jun N-terminalkinase.The human genome includes eight genes that encode BIR-containing

proteins. Among these are cIAP1 and cIAP2, which are highly homol-ogous in amino acid sequence and domain structure, containing (fromthe N to C terminus) three tandem BIR domains followed by a CARDand RING. The cIAP1 and cIAP2 proteins are unique among the BIR-family proteins in their ability to form complexes with TRAF-familyadapter proteins involved in TNF receptor signaling (3–5). Thesemembers of the IAP family are known to induce ubiquitination andproteasome-dependent degradation of the TRAFs to which they bind(e.g. TRAF1, TRAF2), but the structural basis for this phenomenon islargely unknown.In addition to binding certain caspases (6), cIAP1 and cIAP2 are

reported to bind SMAC, a mitochondrial protein that competes withcaspases for binding to IAPs when released into the cytosol (7). In theIAP-familymember, XIAP, the binding site for SMAChas beenmappedto the third BIR domain (BIR3), where it competes for binding tocaspase-9 (8). However, the site on cIAP1 and cIAP2, where SMACbinds has not been previously elucidated, and its relation to the sitesrequired for binding to other targets such as TRAF1 and TRAF2 havenot been heretofore defined.Here, we undertook a structure-function analysis of cIAP1 to com-

pare the binding sites for TRAF2 and SMACand to reveal the functionalconsequences of disruptions in these binding sites. Our findings indi-cate that TRAF2 binds the BIR1 domain, whereas SMACbinds the BIR3domain of cIAP1. The integrity of these binding sites on BIR1 and BIR3is required for cIAP1-mediated ubiquitination of TRAF2 and SMAC,respectively, revealing the basis for differential targeting of protein sub-strates by this E3 ligase. We also show evidence of BIR1-dependentmodulation of TRAF2-mediated regulation of NF-�B. Altogether, thesefindings demonstrate the modularity and diversification of BIRdomains, showing that a single IAP-family protein can direct its E3ligase activity toward different substrates and can alter the cellular func-tions of different protein targets in accordance with differences in thespecificity of individual BIR domains.

MATERIALS AND METHODS

Plasmids, Mutagenesis, Recombinant Proteins—A cDNA cloneencoding cIAP1 (ID 627116) was obtained from IMAGE consortiumthrough ResGen (Invitrogen) and PCR-subcloned into the pcDNA3-FLAG expression vector. Deletion or point mutations were introducedby PCR or using the QuikChange site-directed mutagenesis kit (Strat-agene). All mutations were confirmed by DNA sequencing. The result-ing cDNAs were subcloned into either pGEX4T-1 or a modified

* This work was supported by National Institutes of Health Grants AG15402 (to J. C. R.)and DK067515 (to J. M. Z.). The costs of publication of this article were defrayed in partby 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.

□S The on-line version of this article (available at http://www.jbc.org) contains Supple-mental Figs. S1 and S2.

1 Supported by Department of Defense Postdoctoral Fellowship DAMD17-01-1-0171.2 To whom correspondence should be addressed: Burnham Institute for Medical

Research, 10901 North Torrey Pines Rd., La Jolla, CA 92037. Tel.: 858-646-3140; Fax:858-646-3194, e-mail: [email protected].

3 The abbreviations used are: IAP, inhibitor of apoptosis protein; XIAP, X-chromosome-linked IAP; E3, ubiquitin-protein isopeptide ligase; BIR domain, baculoviral IAP repeatdomain; TNF, tumor necrosis factor; TRAF, TNF receptor-associated factor; HEK cells,human embryonic kidney cells; GST, glutathione S-transferase; UBC, ubiquitin-conju-gating enzyme; SMAC, second mitochondrial activator of caspases; IKK, inhibitor ofNF-�B kinase; RIP, receptor-interacting protein; cFLIP, cellular Fas-associated deathdomain-like interleukin-converting enzyme-inhibitory protein.

THE JOURNAL OF BIOLOGICAL CHEMISTRY VOL. 281, NO. 2, pp. 1080 –1090, January 13, 2006© 2006 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in the U.S.A.

1080 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 281 • NUMBER 2 • JANUARY 13, 2006

by guest on February 5, 2018http://w

ww

.jbc.org/D

ownloaded from

pET21b plasmid. Plasmids encoding some of the cIAP1 proteins and allof the cIAP2, XIAP, and other proteins have been described (6, 9, 10).

Cell Culture and Transfections—Cells were grown in either Dulbec-co’s modified Eagle’s medium (HEK293 and HEK293T) or RPMI(MCF-7) media supplemented with 10% fetal bovine serum, L-gluta-mine, and penicillin/streptomycin. Transfectionswere performed usingSuperfect (Qiagen) for HEK293T cells or Lipofectamine 2000 reagent(Invitrogen) in Opti-MEM-reduced serum medium (Invitrogen). TheLipofectamine transfection mix was replaced with fresh medium after4 h, whereas cells were incubated with Superfect transfection mix for24 h.

Immunoprecipitations—Cells were harvested for SDS-PAGE orimmunoprecipitation �24 h post-transfection. Cells were lysed in ice-cold immunoprecipitation buffer (20 mM Tris-Cl (pH 7.5), 150 mM

NaCl, 10% glycerol, 0. 2% Nonidet P-40, and a protease inhibitor mix-ture (Roche Applied Science). Lysis with immunoprecipitation bufferwas complemented with sonication (5 pulses) on ice. The lysate wascleared of cell debris by centrifugation at 16,000 � g at 4 °C for 30 min.The supernatant was saved for SDS-PAGE or immunoprecipitatedusing anti-FLAGor anti-Myc antibodies conjugated to Sepharose beadsovernight at 4 °C with gentle agitation. The beads were washed 3 timesin lysis buffer, resuspended in 2� Laemmli buffer, and boiled for 5 minto release bound proteins. Proteins were analyzed by SDS-PAGE andimmunoblotting after transfer to polyvinylidene difluoride membranes(Osmonics, Inc.). Competition assays involving immunoprecipitatedproteins were performed by co-incubating the immunoprecipitationreaction with 1 mM competitor tetrapeptides (AVPI, GVPF) or 5 �M

protein (His-BIR3 XIAP). Anti-SMAC antiserum was generated in rab-bits. Other antibodies used were anti-TRAF2 (Santa Cruz, N-19 andC-20), anti-FLAG M2 (Sigma), anti-Myc clone 9E10 (Santa Cruz Bio-technology). AVPI and GVPF peptides were kindly provided by Clem-encia Pinilla (Torrey Pines Institute for Medical Studies) and have beendescribed (11, 12).

Sequence Alignment and HomologyModeling—Sequence alignmentswere performed using BLAST (www.ncbi.nlm.nih.gov/BLAST),DNASTAR, and Genedoc programs. For homology modelingsequences were imported into the First Hand Approach mode of theSwiss Model Automated Comparative Protein Modeling server (13–15). Templates were chosen from the publicly available Protein DataBase (PDB ID 1C9Q, and 1F9X, www.rcsb.org/pdb/index.html) usingthe PDB assigned codes for the solved structures available in the database. Subsequent visualizations, handling, and computations were per-formed using the Deep View Swiss-Pdb Viewer platform (13–15).

In Vitro Protein Interaction and Competitive Binding Assays—GSTfusion proteins were produced in bacteria (Stratagene) usingpGEX4T-1 plasmid containing the cDNA for proteins and affinity-purified using glutathione-Sepharose resin (Amersham Biosciences).Methods used for recombinant protein generation in bacteria and affin-ity chromatography purification have been described (6, 9, 10, 16).L-[35S]Methionine-labeled in vitro translated proteins were producedusing the TNTQuick-Coupled Transcription Translation System (Pro-mega) in 50-�l reactions with 1�g of plasmidDNAper reaction. A totalof 5–10 �l of 35S-labeled in vitro translated mix was incubated withpurified recombinant proteins (20–50 �g) in 0. 3 ml of binding buffer(50 mM Tris-Cl (pH 7.5), 5 mM MgCl2, 5% glycerol, 0. 5 mg/ml bovineserum albumin, and 2mMdithiothreitol, 150mMKCl, and 0. 05%TritonX-100) at 4 °C for 120min. After the incubation, 20�l of binding buffer-equilibrated glutathione-Sepharose beads were added to the mixtureand further incubated for 90min. The beads were washed 3 times for 10min each with 1.0ml of binding buffer. Bound proteins were released by

boiling in 25 �l of SDS sample buffer and analyzed by SDS-PAGE (10%)and autofluorography. For autofluorography, gels were first fixed in a25% isopropanol, 10% acetic acid solution for 20min, briefly rinsed withdeionizedwater, and incubated in a 1M salicylic acid solution for 30min.After a brief washing, the gels were dried and exposed to x-ray film for2–16 h at�70 °C.Where competition assays were performed, the com-petitor peptides (AVPI, GVPF) or proteins (BIR1, SMAC) were co-in-cubated in the binding reaction with the test proteins.

Ex Vivo Ubiquitination Assay—FLAG-tagged cIAP1 proteins wereectopically expressed in 293T cells. After 24 h the cells were lysed inassay buffer (25 mM Tris-Cl (pH 7.5), 50 mM NaCl, 5 mM MgCl2, 1 mM

dithiothreitol, and a protease inhibitor mixture) augmented with briefsonication. The lysates were subjected to immunoprecipitation usinganti-FLAG M2 antibody-conjugated Sepharose beads. Beads werewashed first with 1.0ml of assay buffer for 10min, then oncewith 0. 5mlof assay buffer containing 250mMNaCl for 5min, and a third time with1.0ml of assay buffer for 10min. Thewashed beads were incubatedwitha 26-�l mixture of E1- and E2-containing fractions (8�g of total proteineach), ubiquitin (150 �M), ubiquitin aldehyde (1 �M), and energy-gen-erating system (32 �g) (ubiquitin-protein conjugation kit, Boston Bio-chem). After incubating at 30 °C for 3 h, the reaction was stopped by theaddition of EDTA to a final concentration of 20 mM. The beads werepelleted by centrifugation, and the supernatant was analyzed for forma-tion of high molecular weight ubiquitin conjugate formation by SDS-PAGE and immunoblotting.

NF-�B Reporter Assays—HEK293 or HEK293T cells were seededeither into 96- or 24-well dishes. After 24 h, the cells in 96-well disheswere transfected with 3.3 ng of plasmid encoding Renilla luciferase, 10ng of firefly luciferase, and 8–33 ng of test plasmids per well. For 24-welldishes, each well contained 4 ng of Renilla luciferase, 80 ng of fireflyluciferase, and 600–1000 ng of test plasmids. All transfections wereperformed in triplicate, and the total DNA per well was normalized bythe addition of pcDNA3 empty vector plasmid DNA. Assays resultswere determined 24–27 h after transfection using the Dual LuciferaseAssay system (Promega) according to the manufacturer’s instructions.

RESULTS

The BIR1 Domain of cIAP1 Is Sufficient to Bind TRAF2—Previousstudies have shown that cIAP1 and cIAP2 can bind TRAF1 and TRAF2in the context of TNFRI and TNFRII signaling (3–5). To determine thestructural basis for the interaction of cIAP1 with TRAF2, we expressedvarious fragments of cIAP1 (Fig. 1) as epitope-tagged proteins inHEK293T cells and assessed their ability to interact with endogenousTRAF2 present in the lysates using co-immunoprecipitation assays (Fig.2A). Moreover, by relying on endogenous SMAC present in detergentlysates, we simultaneously assessed the binding of these fragments ofcIAP1 to the SMACprotein. ComparisonsweremadewithXIAP,whichdoes not bind TRAF2 but does bind SMAC (6, 17).

FIGURE 1. IAP proteins and fragments used in the study. Shown is a schematic dia-gram of the cIAP1, cIAP2, and XIAP protein domains and fragments with their respectiveamino acid residues encompassed in the expressed fragments. FL, full-length; BIR, bacu-loviral IAP repeat; CARD, caspase recruitment domain; R, RING domain; d-RING, �-RING;filled rectangles, BIR; filled ovals, CARD; striped rectangles, RING.

E3 Ligase Functions of cIAP1 for TRAF2 and SMAC

JANUARY 13, 2006 • VOLUME 281 • NUMBER 2 JOURNAL OF BIOLOGICAL CHEMISTRY 1081

by guest on February 5, 2018http://w

ww

.jbc.org/D

ownloaded from

TRAF2 co-immunoprecipitated with all fragments of cIAP1 contain-ing the BIR1 domain but failed to associate with all fragments lackingthe BIR1 domain (Fig. 2A). Moreover, TRAF2 co-immunoprecipitatedwith a fragment of cIAP1 containing only the BIR1 domain. In contrast,SMAC co-immunoprecipitated with all cIAP1 fragments that con-tained BIR3 while failing to associate with all cIAP1 fragments lackingBIR3 (Fig. 2A). Unlike BIR3, BIR2 and a fragment containing BIR1–2 ofcIAP1 failed to associate with SMAC (Fig. 2 and supplemental data),suggesting that the BIR1 and BIR2 domains of cIAP1 may not have thestructural determinants required to efficiently bind SMAC in cells.Immunoblot analysis using anti-epitope antibodies confirmed produc-tion of all fragments of cIAP1. Although not as comprehensively ana-lyzed, similar results were obtained for cIAP2, confirming that BIR1 isnecessary and sufficient for co-immunoprecipitating TRAF2 (Fig. 3Band not shown). In contrast to cIAP1, TRAF2 co-immunoprecipitatedneitherwith full-lengthXIAPnorwith any of theXIAP fragments tested(Fig. 2B). However, SMAC did co-immunoprecipitate with full-lengthXIAP and with all fragments of XIAP containing BIR3. SMAC alsoco-immunoprecipitated weakly with fragments of XIAP containingBIR2, consistent with recent comparisons of SMAC-based peptidebinding to BIR2 versus BIR3 of XIAP (18). Thus, cIAP1 and XIAP differin their ability to bind TRAF2 while displaying a conserved ability tobind SMAC via their BIR3 domains.To confirm that the BIR1 is directly responsible for TRAF2 binding,

we performed in vitro protein binding assays using GST fusion proteinsproduced in and purified from bacteria. First, we used the GST pull-downmethod to compare the ability of cIAP1, cIAP2, and XIAP to bindTRAF2 in vitro, employing fragments of cIAP1, cIAP2, and XIAP com-prised of the three tandem BIR domains and using L-[35S]methionine-labeled TRAF2 prepared by in vitro translation (Fig. 3A). TRAF2 asso-ciated with GST-cIAP1 (BIR1–3) and with GST-cIAP2 (BIR1–3) butnot withGST-XIAP (BIR1–3). TRAF2 also did not associate with aGSTfusion protein containing the full-lengthXIAPprotein. Thus, consistentwith prior data, TRAF2 selectively binds cIAP1 and cIAP2, failing toassociate with XIAP. As controls for this TRAF2 binding assay, we also

compared the GST control (non-fusion) protein and GST-CD40 cyto-solic domain, showing binding of TRAF2 to the latter but not the for-mer, consistent with published evidence that TRAF2 binds directly tothe cytosolic tail of CD40 (19).Next, we tested whether GST fusion proteins containing only the

BIR1 domain could bind TRAF2. Because production of the solublerecombinant GST fusion BIR1 domain of cIAP1 proved difficult, weused the BIR1 domain of cIAP2 in an in vitro binding assay. For thisexperiment, we compared GST-BIR1 fusion from cIAP2 with GST-BIR1 of XIAP using in vitro translated 35S-labeled TRAF2. As shown inFig. 3B, the BIR1 domain of cIAP2 bound to TRAF2, whereas BIR1 ofXIAP did not. The binding was specific, as the positive control fragmentcontaining all the three BIR domains of cIAP2, but not the negativecontrol GST protein, was able to bind TRAF2. We also observed thatGST-BIR1 of cIAP2 bound full-length TRAF2 as well as a fragment ofTRAF2 lacking the ring and zinc finger domains but not a fragment ofTRAF2 containing only the C-TRAF domain (supplemental data), con-sistent with prior data indicating that cIAP1 and cIAP2 require thecoiled-coil region of TRAF2 for binding (4).Finally, we used the GST-BIR1 domain of cIAP2 to perform compet-

itive binding experiments as anothermeans of determining whether theBIR1 domain is responsible for TRAF2 binding. For these experiments,lysates were prepared from HEK293T cells that had been transfectedwith plasmids encoding FLAG-cIAP1 and which contain endogenousTRAF2 (and SMAC), then either GST-BIR1 from cIAP2 or GST-BIR1from XIAP protein was added to the lysates as a competitor. FLAG-cIAP1 was immunoprecipitated using anti-FLAG antibody, and associ-ated TRAF2 (and SMAC) were detected by immunoblotting. AlthoughTRAF2 associated with FLAG-cIAP1 immunoprecipitates that wereprepared using lysates that were untreated or that were treated withGST-BIR1 of XIAP, the addition of GST-BIR1 of cIAP2 to lysates inhib-ited TRAF2 binding to FLAG-cIAP1. In contrast, SMAC co-immuno-precipitation with FLAG-cIAP1 was unaffected by the addition of theseGST fusion proteins (Fig. 3C), demonstrating the specificity of the com-petition for TRAF2. After performing immunoprecipitations, glutathi-

FIGURE 2. cIAP1 BIR1 domain, but not XIAP,binds TRAF2. A, the BIR1 domain of cIAP1 is nec-essary and sufficient for interaction with TRAF2.The indicated Myc or FLAG epitope-tagged cIAP1proteins were expressed in 293T cells. Theexpressed proteins were immunoprecipitated (IP)using anti-epitope tag antibodies, and co-immu-noprecipitation of endogenous TRAF2 was analyzedby immunoblotting (WB). The same membraneswere then re-probed with anti-SMAC antibody(fourth panel). The second and fifth panels show theendogenous proteins TRAF2 and SMAC, respec-tively, whereas the third panel shows expression ofthe ectopic IAP and control proteins in lysate.Because of its small size, the cIAP1 BIR3 fragment wasnot resolvable on the same gel, and its expressionwas verified separately (data not shown). The asterisk(*) represents the apparent decrease in SMAC co-im-munoprecipitation due to increased degradationwhen the BIR3-RING fragment is expressed in cells(data not shown). The inset panel (#) shows a longermembrane exposure time to demonstrate BIR2–3cIAP1 expression. FL, full-length. B, XIAP does notbind TRAF2. The indicated XIAP protein fragmentswere expressed in 293T cells. Immunoprecipitationand detection of co-immunoprecipitated proteinswas carried out as in panel A. XIAP-associated pro-teins were detected by immunoblotting using anti-TRAF2 antibody (first panel) or anti-SMAC antibody(fourth panel). The second, third, and fifth panels showthe expression of TRAF2, IAP, and SMAC proteins,respectively. In panels A and B, the protein FLAG-Hippi was expressed as an irrelevant control proteinto demonstrate specificity of protein interactions.

E3 Ligase Functions of cIAP1 for TRAF2 and SMAC

1082 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 281 • NUMBER 2 • JANUARY 13, 2006

by guest on February 5, 2018http://w

ww

.jbc.org/D

ownloaded from

one-Sepharose was added to the same lysates, and the GST fusion pro-teins were recovered and analyzed by SDS-PAGE/immunoblotting forassociated TRAF2. As shown in Fig. 3C, TRAF2 was recovered withGST-BIR1 of cIAP2 but not GST-BIR1 of XIAP, confirming that theBIR1 domain of cIAP2 is sufficient to bind TRAF2. Altogether, thesedata demonstrate the BIR1 domains of cIAP1 and cIAP2 are sufficient tobind TRAF2.

Mutagenesis Analysis of BIR1 Domain of cIAP1 Identifies ResiduesRequired for TRAF2 Binding—TRAF2 binds the BIR1 domain of cIAP1and cIAP2 but not the BIR1 domain of XIAP. We, therefore, comparedthe amino acid sequences of the BIR1 domains of cIAP1 and cIAP2withXIAP to identify differences that might explain their differential abilityto bind TRAF2 (Fig. 4A). Also, because the BIR1 domain of cIAP1 bindsTRAF2, whereas the BIR2 and BIR3 domains of cIAP1 do not, we alsocompared the amino acid sequence of the BIR1 of cIAP1 with the BIR2and BIR3 domains (Fig. 4B). Additionally, to narrow the choice of resi-dues to consider for mutagenesis studies, we employed comparativehomology modeling (13–15) of the cIAP1-BIR1 domain using the pre-viously solved three-dimensional structures for XIAP-BIR2 and XIAP-BIR3 as templates (Fig. 4C). Homology modeling for BIR1 of cIAP1predicted the presence of a typical BIR domain core structure com-prised of two �-helices followed by three antiparallel � strands followedby a third �-helix. We used this model to distinguish residues that werelikely to be surface-exposed from residues buried in the core of thedomain, favoring surface exposed residues for mutagenesis.Because TRAF2 has been reported to bind a peptidyl motif found in

the cytosolic domains of certain TNF-family receptors comprised of theconsensus sequence PXQX(T/S) (20), we focused on regions of the BIR1domain of cIAP1 that showed similarity to this bindingmotif. PXQX(T/S)-like motifs were present at residues 61–65 of cIAP1 (PVSER) and atresidues 99–103 (PIQKH). In both cases these motifs occur at the bor-der of a loop and beginning of an�-helix and are predicted to be surface-

exposed, based on the three-dimensional model. Also, comparison ofthese candidate motifs with the sequences of BIR domains that fail tobind TRAF2 suggested specific residues within these motifs that mightaccount for specificity. For example, the PVSERmotif at residues 61–66of cIAP1 is perfectly matched with the corresponding sequence of theBIR1 domain of cIAP2 but differs from XIAP, which has the sequencePVSAS (Fig. 4A, motif 1). Also, the PIQKH motif at 99–103 of cIAP1compares with the similar sequence of PTEKH in cIAP2 but is quitedifferent from the corresponding AVGRH sequence within BIR1 ofXIAP (Fig. 4,motif 2). Additional sequences resembling the PXQX(T/S)motif were also found in BIR1 of cIAP1 but were discounted as likelybinding sites for TRAF2 because of their absence in the BIR1 sequenceof cIAP2 or presence in the sequences of non-TRAF binding BIRs orbecause they were not predicted to be surface-exposed.We engineered alanine substitutions within motif 1 of S63A, E64A,

and the double mutant E64A/R65A. In addition, we made non-conser-vative substitutions of E64R andR65E. Formotif 2, we generated alaninesubstitution mutants of P99A, Q101A, and the double mutant P99A/Q101A. These mutant BIR1 proteins were then expressed as Myc-tagged proteins in HEK293T cells and assessed for their ability to bindendogenous TRAF2 present in the cell lysates by co-immunoprecipita-tion assay. As shown in Fig. 4D, alanine replacements withinmotif 2 didnot impair TRAF2 binding. In contrast, all substitutions in motif 1,except for S63A, ablated TRAF2 binding, including the E64A, E64R,R65E, and E64A/R65A mutations.To verify that the E64 and R65mutations directly affect BIR1-TRAF2

association, we used cIAP2 instead of cIAP1, and we produced recom-binant single mutant E64R and double mutant E64A/R65A BIR1 pro-teins in bacteria and examined their interaction with TRAF2 in vitro.Similar to our data from in vivo experiments using cell lysates (see Figs.4 and 7), E64R mutant GST-BIR1 weakly bound TRAF2 in vitro,whereas the double mutant E64A/R65A did not bind TRAF2 (Fig. 5A).

FIGURE 3. cIAP1 and cIAP2 interact with TRAF2.A, both cIAP1 and cIAP2 interact with TRAF2. Theinteraction of the indicated recombinant GSTfusion proteins (BIR1–3 domains of cIAP1, cIAP2,and XIAP, full-length XIAP and CD40, or GST-only)with in vitro translated, [35S]methionine-labeledTRAF2 was examined by GST pull-down assay.Bound proteins were resolved by SDS-PAGE anddetected by autofluorography. The lower panelshows a Coomassie-stained gel of the recombi-nant proteins with the relative position of themolecular mass markers indicated (kDa). Arrowsindicate positions of the recombinant proteins. B,the BIR1 domain of cIAP2 is sufficient to bindTRAF2. GST fusion BIR1 domains of cIAP2 and XIAPwere analyzed for their interaction with in vitrotranslated [35S]methionine-labeled TRAF2. Bind-ing was detected by autofluorography as in panelA. The lower panel shows Coomassie-stainedrecombinant proteins used for the assay. C, BIR1domains of cIAP1 and cIAP2 bind TRAF2 in an iden-tical manner. FLAG-cIAP1 was overexpressed in293T cells. The cells were then lysed, and cIAP1was immunoprecipitated (IP) using beads conju-gated to anti-FLAG antibody in the presence orabsence of the indicated GST fusion BIR1 compet-itors (10 and 30 �M). FLAG beads were pelleted,and the resulting supernatant was incubated withglutathione beads to pull down the GST fusioncIAP2 BIR1. TRAF2 binding to FLAG-cIAP1 (firstpanel) or GST- cIAP2 BIR1 (second panel) was exam-ined by SDS-PAGE and immunoblotting (IB) withantibodies against TRAF2. The membrane was alsoprobed with an anti-SMAC antibody (third panel)to examine if BIR1 could interfere with cIAP1-SMAC binding. Finally, the membrane was stainedwith Coomassie to detect the GST-BIR1 proteinsused for the competition (fourth panel).

E3 Ligase Functions of cIAP1 for TRAF2 and SMAC

JANUARY 13, 2006 • VOLUME 281 • NUMBER 2 JOURNAL OF BIOLOGICAL CHEMISTRY 1083

by guest on February 5, 2018http://w

ww

.jbc.org/D

ownloaded from

Additionally, we performed a competition assay to examine if themutant BIR1 proteins would displace wild-type BIR1 from TRAF2. Asshown in Fig. 5B, high concentrations of wild-type BIR1, but not theE64R or E64A/R65Amutants, efficiently displaced wild-type GST-BIR1from binding TRAF2. To exclude the possibility that the mutations weintroduced to BIR1 may have caused unfolding of the domain, therebynon-specifically disrupting the binding to TRAF2, we produced therecombinant mutant proteins and compared them with the wild-typeBIR1 protein by size exclusion chromatography and one-dimensionalNMR analysis. The mutants and wild-type BIR1 proteins had identicalchromatographic retention profiles (Fig. 5C) and one-dimensionalNMR spectra (not shown), ruling out protein unfolding. We conclude,therefore, that motif 1 is the likely binding site for TRAF2 on BIR1 ofcIAPs.

Mutational Analysis of the SMAC-binding Site on cIAP1—Previously,a SMAC-binding site has been delineated on the BIR3 domain of XIAP

which consists of a groove that accommodates the first four amino acidsof the N terminus of mature SMAC having the sequence AVPI (8, 17,18). Homology modeling of the BIR3 domain of cIAP1 suggests a verysimilar groove (Fig. 6A). We, therefore, performed mutagenesis studiesof the candidate SMAC-binding site on BIR3 of cIAP1, modeling ourapproach after prior work on XIAP (17). Accordingly, alanine substitu-tion mutations were made at Asp-320 and Trp-316 that were predictedto be involved in SMAC tetrapeptide binding based on comparisonswith XIAP. By analogy to XIAP, Asp-320 of cIAP1 is predicted to forman electrostatic interaction with the alanine of the AVPI N terminus ofSMAC, whereas Trp-316 is predicted to give structural stability to theSMAC groove and form a hydrophobic pocket for the AVPI tetrapep-tide (17). As controls, mutations were also made at Ser-129, located atthe end of BIR1, a site predicted to have no effects, and at Asp-234 inBIR2 of cIAP1, in a region predicted to be analogous to the groove ofBIR3 that binds SMAC.

FIGURE 4. A BIR1-specific sequence is essential for BIR1-TRAF2 interaction. A, sequence alignment of the BIR1 domains of cIAP1, cIAP2, and XIAP. Predicted �-helical (cylinders)and �-strand (arrows) structures on cIAP1 BIR1 are shown above the aligned sequence. B, sequence alignment of the BIR1, BIR2, and BIR3 domains of cIAP1. Motif 1 and motif 2 in A andB indicate the positions of PXQX(T/S)-like motifs in the BIR1 amino acid sequence of IAPs. Mutations were created at these two sites to examine the effect on the BIR1-TRAF2interaction. Residues showing sequence similarity among the BIR1 sequences are boxed. C, homology modeling and molecular surface predictions for cIAP1 BIR1. A ribbon diagram(left) and electrostatic charge and molecular surface (right) models for cIAP1 BIR1 are shown. The predicted side chain orientations for residues Glu-64, Arg-65, Pro-99, and Gln-101 arealso shown. D, residues Glu-64 and Arg-65 are critical for cIAP1-TRAF2 interaction. The indicated Myc epitope-tagged wild-type (wt) or mutant BIR1 proteins were expressed in 293Tcells. The overexpressed proteins were immunoprecipitated (IP) with anti-Myc antibodies, and co-immunoprecipitation of endogenous TRAF2 was analyzed by immunoblotting withanti-TRAF2 (N-19) antibodies. The reason for mobility shift in the expressed BIR1 proteins is unclear.

E3 Ligase Functions of cIAP1 for TRAF2 and SMAC

1084 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 281 • NUMBER 2 • JANUARY 13, 2006

by guest on February 5, 2018http://w

ww

.jbc.org/D

ownloaded from

For these experiments, wild-type cIAP1 and mutants of cIAP1 wereexpressed with FLAG tags in HEK293T cells, and their ability to bindmature SMAC protein present within detergent lysates was analyzed bySDS-PAGE/immunoblotting. As shown in Fig. 6B, the W316A mutantof cIAP1 was markedly deficient in SMAC binding activity, whereas theD320A mutant showed some reduction in SMAC binding. In contrast,the BIR1 mutant (S129A) and BIR2 mutant (D234A) retained fullSMAC binding activity, thus confirming the specificity of the BIR3mutations. All cIAP1mutants were expressed at levels comparable withor greater than wild-type cIAP1, as determined by immunoblot analysisof the cell lysates (Fig. 6B), thus excluding differences in cIAP1 proteinproduction as an explanation for the differential binding to SMAC.Equivalent levels of endogenous SMAC were also found in the lysatesderived from all transfected cell samples (Fig. 6B).Next, to confirm the specificity of the mutagenesis results, we com-

pared the binding of cIAP1mutants in which either the TRAF2-bindingsite within BIR1 had been altered or the SMAC-binding site within BIR3had been mutated (Fig. 6C). Again, co-immunoprecipitation assayswere employed for measuring protein interactions, except we reversedthe immunoprecipitating antibody and used anti-SMAC to recover pro-tein complexes as a further confirmation of the results. As shown in Fig.6C, full-length cIAP1 and cIAP1(E64A/R65A), the TRAF2 bindingmutant, co-immunoprecipitated with SMAC. In contrast, the BIR3mutants cIAP1(W316A) and cIAP1(D320A) did not. Immunoblot anal-ysis confirmed the presence of comparable amounts of SMAC proteinand of FLAG-tagged cIAPwild-type andmutant proteins in all samples,thus excluding differences in input amounts of proteins as a trivialexplanation for the findings (Fig. 6C).

To further corroborate the conservation of structural determinantsof the SMAC binding pockets within the BIR3 domains of cIAP1 andXIAP, we compared the ability of SMAC tetrapeptides to compete forbinding of cIAP1 and XIAP to mature SMAC protein using co-immu-noprecipitation assays. For these experiments, full-length cIAP1 andXIAP were expressed in HEK293T cells as FLAG-tagged proteins, thencell lysates were prepared to which SMAC tetrapeptide AVPI wasadded. Lysates were also treated with a negative control tetrapeptide,GVPF, that does not bind the BIR3 domain of XIAP, or with GST-BIR3protein as a positive control for competition. Then FLAG-cIAP1 andFLAG-XIAP were recovered from lysates by immunoprecipitation, and

bound SMAC was detected by immunoblotting. As shown in Fig. 6D,AVPI peptide effectively blocked binding of both cIAP1 and XIAP toSMAC protein. In contrast, the control peptide GVPF did not interferewith cIAP1 and XIAP binding to SMAC protein. The BIR3 fragment ofXIAP partially displaced full-length cIAP1 and XIAP from SMAC, asexpected. Immunoblot analysis confirmed equivalent total levels ofSMAC protein in the cell lysates and production of comparableamounts of FLAG-cIAP1 and FLAG-XIAP (Fig. 6D), thus excludingdifferences in input amounts of proteins as an explanation for the find-ings. The inhibition of SMAC/cIAP1 binding by SMAC peptide AVPIwas specific in that binding of TRAF2 to BIR1 of cIAP2 was not sup-pressed (Fig. 6E). We conclude, therefore, that the BIR3 domains ofcIAP1 and XIAP are conserved in their modes of SMAC binding.

Analysis of RING DomainMutants of cIAP1—The RING domains ofIAP-family protein are believed to bind ubiquitin-conjugating enzymes(UBCs), thus accounting for the E3 ligase activity of these proteins. As aprelude to undertaking functional studies of the E3 ligase activity ofcIAP1, we determined the effects of deleting the RING domain or ofmutating a conserved histidine residue (His-588) predicted to berequired for coordinating zinc and, thus, maintaining integrity of theRING domain. Fragments of cIAP1 lacking the RING domain and thecIAP1(H588A) mutant retained their ability to bind TRAF2 and SMAC(Fig. 7, A and B), indicating the deletion or disruption of the RINGdomain does not impair TRAF2- and SMAC binding activity of cIAP1.In contrast, the aforementioned mutations in BIR1 and BIR3 inhibitedbinding of cIAP1 to TRAF2 and SMAC, respectively, thus serving asspecificity controls.

Ubiquitination of SMAC and TRAF2 by cIAP1 Depends on BIR-me-diated Binding and RING Integrity—We performed a structure-func-tion analysis of the E3 ligase activity of cIAP1 with respect to its sub-strates TRAF2 and SMAC. To this end, we performed ex vivoubiquitination assays using FLAG-tagged wild-type and mutant cIAP1proteins recovered by immunoprecipitation using anti-FLAG-Sepha-rose and then incubated with ubiquitination-competent reaction mixcontaining E1/E2 and other components. The status of ubiquitinationof associated TRAF2 and SMAC was analyzed by SDS-PAGE/im-munoblotting. Comparisons were made of 1) wild-type cIAP1, 2)cIAP1(BIR1–3) lacking the RING domain, 3) cIAP1(E64A/R65A),

FIGURE 5. E64 and R65 mutations of cIAP2 BIR1disrupt interaction with TRAF2 in vitro. A, inter-action of GST-tagged wild-type (wt), E64R, orE64A/R65A (ER-AA) recombinant cIAP2 BIR1 pro-teins with 35S-labeled in vitro translated TRAF2 wasanalyzed by GST pull-down assay. RecombinantGST protein was used as a negative control. Theupper panel shows 35S-labeled TRAF2 autofluoro-graph, whereas the lower panel shows a Coomassie-stained gel of the recombinant proteins used (4�g/lane). B, E64 and R65 mutant BIR1 proteins do notdisrupt the association of wild-type BIR1 with TRAF2.GST-tagged BIR1–3 protein (5 �M) was allowed toassociate with 35S-labeled in vitro translated TRAF2 inthe presence or absence of 10 or 50 �M competitoruntagged BIR1 (wild-type, E64R, E64A/R65A) pro-teins. TRAF2 bound to the GST-tagged BIR1–3 wasexamined by pull-down assay followed by autoflu-orography. SMAC protein was used as a negativecontrol. C, wild-type, and TRAF2 binding mutantE64R and E64A/R65A recombinant BIR1 proteinswere compared by gel-sieve chromatography. The xaxis shows the relative retention volume of the pro-teins in Superdex 200 column. AU, absorbance units.D, preparations of untagged BIR domains used in Band C were analyzed by SDS-PAGE Coomassie stain-ing (4 �g). Relative molecular mass markers (in kilo-daltons) are indicated.

E3 Ligase Functions of cIAP1 for TRAF2 and SMAC

JANUARY 13, 2006 • VOLUME 281 • NUMBER 2 JOURNAL OF BIOLOGICAL CHEMISTRY 1085

by guest on February 5, 2018http://w

ww

.jbc.org/D

ownloaded from

which fails to bind TRAF2, 4) cIAP1(D320A), with reduced SMACbinding activity, and 5) cIAP1(H588A), in which the RING is disturbed.High molecular weight conjugates of SMAC were produced by wild-

type cIAP1 and cIAP1(E64A/R65A) but not by cIAP1(BIR1–3),cIAP1(D320A), or cIAP1(H588A) (Fig. 8). Co-immunoprecipitationassays showed that cIAP1(BIR1–3) and cIAP1(H588A) bound SMACbut did not promote its modification, whereas cIAP1(D320A) neitherbound nor promotedmodification of SMAC.Thus, for cIAP1-mediatedubiquitin conjugation of SMAC, an intact BIR3 domain and RINGdomain are both required.Interpretation of results for TRAF2 was somewhat complicated by

higher background rates of ubiquitin conjugation, possibly attributableto TRAF2 ability to bind Ubc13 via its RING domain (21). However,

wild-type cIAP1 and cIAP1(D320A) clearly increased the amounts ofhigher molecular weight conjugates of TRAF2 compared withcIAP1(BIR1–3) and cIAP1(H588A) (Fig. 8). Co-immunoprecipitationassays showed that binding of TRAF2 to cIAP1(E64A/R65A) wasmark-edly reduced compared with other cIAP1 proteins tested, as expected,whereas cIAP1, cIAP1(D320A), cIAP1(BIR1–3), and cIAP1(H588A)bound TRAF2 to comparable extents. Together, these data suggest thatan intact BIR1 domain and RING domain of cIAP1 are required forbinding and modification, respectively, of TRAF2.

Modulation of TRAF2-induced NF-�B Activation by cIAP1,Mutagenesis Analysis—TRAF2 is an adaptor molecule known to regu-late NF-�B and Jun N-terminal kinase signaling pathways. Overexpres-sion of TRAF2 induces NF-�B activation (22), providing a convenient

FIGURE 6. SMAC binds to the BIR3 domains of cIAP1 and XIAP in a conserved manner. A, cIAP1 and XIAP residues critical for SMAC binding are conserved. A ribbon diagram modelof cIAP1 BIR3 is shown in comparison with the ribbon diagram derived from the structure of XIAP BIR3. The side chains of cIAP1 BIR3 residues Asp-320 and Trp-310 are shown againstthe corresponding E314 and Trp-310 residues of XIAP BIR3. B, FLAG-tagged cIAP1 full-length proteins mutated in BIR2 domain (D234A), BIR3 domain (D320A, W316A), or at the endof BIR1 (S129A) were expressed in 293T cells, and the interaction of the expressed proteins with SMAC was determined by immunoprecipitation (IP) using anti-FLAG followed byimmunoblotting (IB) using anti-SMAC antibodies (top panel). Empty pcDNA3 or pcDNA3 encoding wild-type (wt) BIR domains were transfected as negative and positive controls,respectively. The lower two panels show the expression levels of cIAP1 proteins and endogenous SMAC. C, mutated cIAP1 BIR3 does not interact with SMAC. FLAG-tagged BIR3 W316Aor BIR3 D320A, BIR1 E64A/R65A TRAF binding mutant, and BIR3-RING cIAP1 proteins were expressed in 293T cells. After 24 h, endogenous SMAC was immunoprecipitated with theanti-SMAC antibody, and bound FLAG-cIAP1 proteins were detected with anti-FLAG antibodies by immunoblotting (first panel). Joined arrows indicate the positions of the FLAG-tagged full-length cIAP1 proteins and the BIR3-RING fragment positions. Cleaved fragments of E64A/R65A and wild-type cIAP1 proteins are also detected in the respective lanes.SMAC and FLAG-cIAP1 protein levels in the lysate are shown in the lower two panels. D, the IAP binding motif peptide of SMAC interferes with the interaction between cIAP1 andSMAC. FLAG-tagged wild-type cIAP1 and XIAP proteins were expressed in 293T cells. Expressed IAP proteins were immunoprecipitated with anti-FLAG antibodies in the presence ofAVPI tetrapeptide derived from the SMAC IBM motif, GVPF control tetrapeptide (both peptides at 1 mM final concentration), or His-tagged BIR3-XIAP (at 5 �M final concentration) ascompetitors. Co-purified endogenous SMAC protein was detected using anti-SMAC antibody (top panel). The lower two panels show expression of SMAC and IAP proteins in thelysates used for the competition assay. E, SMAC does not disrupt in vitro BIR1-TRAF2 association. In vitro interaction (IVT) between GST fusion cIAP2-BIR1 (5 �M) and 35S-labeled TRAF2was examined by GST pull-down assay in the presence of the indicated competitor peptides (AVPI and GVPF, both at 500 �M final concentration) or proteins (SMAC and BIR1, bothat 50 �M final concentration). Me2SO (DMSO) was used as carrier control for the peptides. Bound 35S-labeled TRAF2 was detected by autofluorography as described under “Materialsand Methods.” Two micrograms of the recombinant proteins was resolved by SDS-PAGE and Coomassie-stained to show that the recombinant proteins were expressed at a similarabundance (right panel).

E3 Ligase Functions of cIAP1 for TRAF2 and SMAC

1086 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 281 • NUMBER 2 • JANUARY 13, 2006

by guest on February 5, 2018http://w

ww

.jbc.org/D

ownloaded from

endpoint for assessing the effects of cIAP1 on TRAF2 activity. Weassessed the functional consequences of cIAP-TRAF2 interaction byusing NF-�B reporter assays. To this end, we first assayed the effects onNF-�B reporter gene activity of expressing TRAF2, wild-type cIAP1, orcIAP1mutants (BIR1–3, BIR3-RING, H588A) in HEK293T cells. Com-parisons were alsomadewithXIAP. As shown in Fig. 9A, transfection ofincreasing amounts of TRAF2-encoding plasmid DNA induced dose-dependent increases inNF-�B reporter gene activity. XIAP also inducedslight increases inNF-�B activity. In contrast, cIAP1 and cIAP1mutantsdid not induce NF-�B.Next, we compared the effects of co-expressing TRAF2 with wild-

type cIAP1 or cIAP1 mutants. Full-length, wild-type cIAP1 had no sig-nificant effect on TRAF2-mediated induction of NF-�B activity (Fig.9B). Similarly, cIAP1mutants withmodified BIR1 and BIR3 domain hadno effect. However, TRAF2-mediated induction of NF-�B activity wassignificantly elevated by co-expression of cIAP1(BIR1–3), which lacksthe RING domain, or cIAP1(H588A), in which the RING domain hasbeen perturbed.Next, we attempted to dissect the elements within the BIR1–3 frag-

ment of cIAP1 required for enhancing TRAF2-mediated NF-�B activ-ity. To this end we tested cIAP1 deletion mutants encoding only BIR1,BIR1–2, or BIR1–3 fragments in TRAF2/NF-�B reporter assay.We alsointroduced mutations into BIR1 and BIR3 of the cIAP1(BIR1–3) frag-ment that inhibit binding to TRAF2(E64A/R65A) and SMAC(D320A),respectively, and tested the ability of these mutant BIR1–3 fragments ofcIAP1 to enhanceNF-�B induction byTRAF2.Mutation of the TRAF2-binding site in the BIR1 domain completely abrogated enhancement ofTRAF2-mediated NF-�B induction (Fig. 9C). In contrast, mutation ofthe SMAC-binding site on BIR3 partially inhibited the ability of theBIR1–3 fragment of cIAP1 to enhance TRAF2-induced NF-�B activity.These observations suggest that the TRAF2-binding site on BIR1 isabsolutely required for NF-�B-enhancing activity but also suggest thatthe SMAC-binding site on BIR3 may make contributions. It remains tobe determined whether SMAC versus other proteins that are known tobind BIR3 participates in the enhancement of NF-�B activation.

To explore whether the ability of RING-less cIAP1 to enhanceTRAF2-mediated NF-�B induction is unique to cIAP, we compared theeffect of fragments of cIAP1, cIAP2, and XIAP that contained only thethree tandem BIR domains. Both cIAP1(BIR1–3) and cIAP2(BIR1–3)substantially enhanced TRAF2-mediated NF-�B activity, whereasXIAP(BIR1–3) did not (Fig. 9D). These data are consistent with theability of cIAP1 and cIAP2, but not XIAP, to bind TRAF2 via their BIR1domains.Finally, we examined if the RING domain of TRAF2 is essential for

enhancement of NF-�B activation by the BIR1–3 fragment of cIAP1.For these experiments we compared wild-type TRAF2 with a TRAF2deletion mutant (N-terminal amino acids 1–87 deleted (dN)) and witha RING domain site-specific mutant (C49A/H51A double mutant) ofTRAF2 proteins in co-transfection reporter gene assays where theseproteins were co-expressed with BIR1–3 of cIAP1. As shown in Fig. 9E,whereas BIR1–3 clearly enhanced the NF-�B activity induced by wild-type TRAF2, the RING mutants of TRAF2 neither induced NF-�B nordid BIR1–3 influence their activity. Therefore, we conclude thatTRAF2-mediated induction of NF-�B activity is dependent on theRING domain TRAF2 and that cIAP(BIR1–3) cannot overcome therequirement for the RING of TRAF2.

DISCUSSION

Although IAPs share structurally conserved BIR domains, individualmembers of the family have acquired distinct functions, subcellularlocalizations, and the ability to interact with different proteins. The

FIGURE 8. cIAP1 ubiquitinates TRAF2 and SMAC through distinct binding domains.Control pcDNA3 vector or cIAP1 mutant plasmids (wt, E64A/R65A TRAF2 binding mutant(ER-AA), D320A SMAC binding mutant, BIR1–3, and H588A RING-mutated) were trans-fected into 293T cells. After 24 h cIAP1 protein complexes were affinity-purified usinganti-FLAG antibody-coated beads. An ex vivo ubiquitination assay was performed byincubating the cIAP1-complexed proteins with exogenous ubiquitin, E1/E2-containingfractions, and an energy-generating system as described under “Materials and Meth-ods.” Ubiquitination of SMAC and TRAF2 was analyzed by SDS-PAGE of the reactionmixture and immunoblotting with anti-SMAC and anti-TRAF2 antibodies (first and thirdpanels). Levels of SMAC, TRAF2, and expressed cIAP1 proteins in the lysate are shown insecond, fourth, and fifth panels. The relative molecular mass values are shown (kDa).

FIGURE 7. RING mutation does not affect interaction of cIAP1 with TRAF2 or SMAC.A, FLAG-tagged TRAF binding mutant E64R, RING mutant H588A, and deletion mutant(BIR1–3 and BIR3-RING) cIAP1 proteins were expressed in breast carcinoma MCF-7 cells.After 24 h, endogenous TRAF2 was immunoprecipitated (IP), and the co-purified FLAG-cIAP1 proteins were detected by immunoblotting (IB) with anti-FLAG antibodies (upperpanel). The lower panel shows cIAP1 protein expression. wt, wild type. B, FLAG-taggedcIAP1 proteins mutated at BIR1 (P99A or E64A) or RING-domain (H588A) were expressedin 293T cells. The expressed FLAG-cIAP1 proteins were purified by immunoprecipitation,and co-purified TRAF2 and SMAC proteins were detected by immunoblotting (top twopanels). The lower three panels indicate the levels of TRAF2, SMAC, and FLAG-cIAP1 pro-teins in the lysate.

E3 Ligase Functions of cIAP1 for TRAF2 and SMAC

JANUARY 13, 2006 • VOLUME 281 • NUMBER 2 JOURNAL OF BIOLOGICAL CHEMISTRY 1087

by guest on February 5, 2018http://w

ww

.jbc.org/D

ownloaded from

proteins cIAP1 and cIAP2 consist of three tandem BIR domains, aCARDdomain, and a RING-finger domain. These IAP-familymemberswere first identified by virtue of their ability to associate with TNF-family receptors via associations with TRAF1 and TRAF2. Since then,several other proteins have been reported also to interact with cIAP1 orcIAP2, including caspases-3, -7, and -9, SMAC, HtrA2, IKK�, and RIP(6, 7, 23–25). The structural determinants involved in interactions ofcIAP1 and cIAP2 for most of these protein interactions are presentlyunknown. A detailed understanding of how cIAP1 and cIAP2 interactwith various cellular proteins could be useful both for improving under-standing about the mechanisms of these proteins and also for poten-tially generating chemical inhibitors that modulate their activity intherapeutically useful ways. We, therefore, sought to identify theTRAF-binding site on cIAP1, making comparisons with the SMAC-binding site.We determined that the BIR1 domain of cIAP1 and cIAP2 is neces-

sary and sufficient for binding TRAF2. TRAF-family proteins, includingTRAF2, contain a surface crevice that binds peptidyl motifs foundwithin the cytosolic domains of certain TNF-family receptors andadapter proteins (26, 27). Within BIR1 of cIAP1 and cIAP2 are twoamino acid sequence motifs resembling the known peptidyl ligands of

TRAFs. Bymutagenesis studies we determined that amotif predicted tobe located in a loop that extends into the second �-helix of the BIR1domain appears to be required for TRAF2 binding. The mutations wegenerated did not disrupt the overall folding of BIR1, suggesting that theidentified motif is directly involved in TRAF2 binding. Alternatively, anon-linear interaction module analogous to SMAC-BIR3 binding,which would engage the N-terminal loop and helices of cIAP BIR1 withthe coiled coil segment of TRAF2, is conceivable. However, these dataderived from mutagenesis must be confirmed by structural analysis orother biophysical approaches to confirm direct involvement of thePXXER motif at residues 61–65 in TRAF2 binding. Amino acidsequence alignments ofmammalian BIR domain reveals conservation ofthe PXXER motif only in the BIR1 domains of cIAP1 and cIAP2, con-sistentwith previously published observations that cIAP1 and cIAP1 areunique among mammalian IAPs in their ability to bind TRAFs.In contrast to TRAF2, we determined that the BIR3 domain of cIAP1

is necessary and sufficient for binding SMAC. Several mammalian IAP-family proteins have been reported to bind SMAC, including XIAP,cIAP1, cIAP2, and ML-IAP. The structural basis for the interactionof SMAC with IAP-family proteins has been determined for XIAP,where NMR, x-ray crystallographic, and other supporting studies have

FIGURE 9. cIAP-BIR domains modulate TRAF2-induced NF-�B activation. A, TRAF2 or the indicated wild-type (wt) or mutant IAP proteins were expressed, and NF-�B reporteractivity was assayed as described under “Materials and Methods.” B, loss of cIAP1 E3-ligase activity promotes NF-�B activation by TRAF2. TRAF2 was transfected alone or incombination with the indicated cIAP1 mutants (wild-type, E64A/R65A, D320A, BIR1–3 domain, H588A). After 24 h NF-�B reporter activity was assayed. C, both BIR1 and BIR3 arerequired for enhancement of NF-�B activation. TRAF2 was co-transfected with pcDNA3 vector, BIR1, BIR1–2, BIR1–3, TRAF binding mutant BIR1–3 (ER-AA), or SMAC-pocket mutantBIR1–3(D320A). NF-�B activity was assayed after 24 h. D, cIAP1 and cIAP2 both enhance NF-�B activity induced by TRAF2 through their BIR1–3 domains. TRAF2 was co-transfectedwith pcDNA3 vector, cIAP1 BIR1–3, cIAP2 BIR1–3, or XIAP BIR1–3, and the NF-�B reporter activity was assayed. E, the RING domain of TRAF2 is required for BIR enhancement of NF-�Bactivity. cIAP1 BIR1–3 fragment protein was co-expressed with wild-type TRF2 (wt), a TRAF2 mutant that lacks the N-terminal 86 amino acids containing the RING domain (dN), or aC49A/H51A RING point-mutant TRAF2 (RM). NF-�B reporter activity was assayed 24 h after transfection. Representative graphs from at least two independent experiments were runin triplicate are shown. Error bars represent S.D.

E3 Ligase Functions of cIAP1 for TRAF2 and SMAC

1088 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 281 • NUMBER 2 • JANUARY 13, 2006

by guest on February 5, 2018http://w

ww

.jbc.org/D

ownloaded from

revealed a tetrapeptide motif in the N terminus of proteolytically pro-cessed SMAC that binds a surface crevice on the BIR3 domain of XIAP.Homology modeling and sequence analysis suggest that the BIR3domain of cIAP1 is very similar toXIAP, with conservation of 8 of the 11residues in XIAP that have been determined to account for interactionswith the N terminus of mature SMAC. Accordingly, our mutagenesisanalysis of the putative SMAC binding pocket on BIR3 of cIAP1 pro-vides additional supportive evidence that themode of SMACbinding tocIAP1 is nearly identical to XIAP.The presence of a UBC binding RING domain in cIAP1 and cIAP2

supports data indicating that these proteins function, at least in part, asE3 ligases. In previous reports it was shown that TRAF2 is ubiquitinatedby cIAP1 and that the RINGdomain of cIAP1, but not the RINGdomainof TRAF2, is required for the ubiquitination (5). We show here thatfull-length wild-type cIAP1 binds and induces ubiquitination of TRAF2in vitro. Disruption of the TRAF2-binding site on BIR1 or disruption (ordeletion) of the RING domain of cIAP1 ablates binding and ubiquitina-tion, respectively, indicating that both the BIR1 and RING domains arerequired for TRAF2 ubiquitination. In contrast, we demonstrated dif-ferent requirements for binding and ubiquitination of SMAC, where anintact BIR3 domain plus intact RING domain are required for bindingand ubiquitination of SMAC. The distinct and independent require-ments of BIR1 and BIR3 for binding TRAF2 and SMAC, respectively,demonstrate the modularity of cIAP1 as an E3 ligase, illustrating howdifferent BIR domains are used to diversify substrates targeted for ubiq-uitination. It will be interesting in future studies to delineate the bindingsites of other proteins reported to bind cIAP1 to compare their interac-tion sites with TRAF2 and SMAC.During attempts to study the effects of cIAP1 on TRAF2 function

with respect to NF-�B activation, we noticed that expression of a frag-ment of cIAP1 consisting of the three BIR domains greatly enhancesNF-�B activity induced by overexpression of TRAF2. Disruption of theRINGdomain of cIAP1 by aH588Amutation partially but incompletelyrevealed this phenotype of cIAP1, implying that structures downstreamof the BIR domains somehow suppress NF-�B enhancement. A BIR1–3fragment of cIAP2 also enhanced TRAF2-mediated NF-�B activity,whereas BIR1–3 of XIAP did not, implying that TRAF2 binding isrequired. Consistent with a requirement for TRAF2 binding, mutationsin BIR1 that prevent TRAF2 binding abrogated the NF-�B-enhancingeffect of cIAP1 (BIR1–3).We presume the ability of BIR1-BIR3 fragments of cIAP1 and cIAP2

to enhance TRAF2-induced NF-�B activity is attributable to competi-tion with endogenous cIAP1 or cIAP2, possibly protecting TRAF2 fromubiquitination in cells. Thus, it is possible that cIAP1 and cIAP2 mayprovide a negative feedback function to check excessive signaling byTNF-family receptors. However, we have been unable to demonstrate areduction in TRAF2 levels in cells transfected with cIAP1 or cIAP2,suggesting the effects on TRAF2 degradation may only partly accountfor the phenomenon that BIR1–3 fragments of cIAP1 and cIAP2enhance NF-�B induction by TRAF2. For example, recent studies alsosuggest that ubiquitination of TRAF2 can alter its subcellular localiza-tion, which could impact its ability to induce NF-�B (21, 28). Preciselyhow IAPs modulate the TNF-NF-�B pathway is not completely under-stood. Among the reported targets of cIAP1 and cIAP2 that could con-ceivably be involved are IKK�, cFLIP, RIP, andTRAF2.Also, whereaswedo not know the reason that cIAP1 and cIAP2 did not cause reductionsin TRAF2 in intact cells, it has been speculated that competitionbetween autoubiquitination of TRAF2 with the Lys-63-linked ubiquitinchain (due to association of its RING with Ubc13) with Lys-48-linked

ubiquitin chains (coming from cIAP1/2) may prevent proteasome-de-pendent degradation (29).Although the underlying mechanisms accounting for enhanced

TRAF2-mediated NF-�B activation by BIR1–3 fragments of cIAP1 andcIAP2 remain to be defined, the observations made here are likely tohave relevance to the pathogenesis of mucosa associated lymphoidtumors (MALTomas), a type of non-Hodgkin lymphomas, where fusionof the cIAP2 and MALT1 genes occurs as a result of chromosomaltranslocations (30). These translocations create cIAP2/MALT1 fusionproteins adjoining theN-terminal portion of cIAP1 containing the threeBIRs to a portion of MALT1, thus separating the BIR1–3 region fromthe rest of the cIAP2 molecule. Expression of cIAP2/MALT1 fusionprotein inducesNF-�B activation, an event linked to promoting survivaland proliferation of B-cells and, thus, presumably of relevance to theoncogenic mechanism. Previously, it was demonstrated that deletion ofBIR1 from cIAP2/MALT1 fusion protein abrogated NF-�B-inducingactivity, whereas deletion of a putative TRAF6-binding site on MALT1was inconsequential (31). The role of BIR1 was attributed to dimeriza-tion/oligomerization, thus causing activation of MALT1. However,becausewe showhere that BIR1 is required for TRAF2 binding, it will beinteresting in the future to ascertain the effects on NF-�B of mutationsin cIAP2/MALT1 fusions that preclude binding of TRAF2. In thisregard it could be that BIR1-mediated dimerization/oligomerization isactually due to binding TRAF2, which is a trimeric molecule. Agentsthat disrupt binding of TRAF2 to cIAP2/MALT1 fusion proteins con-ceivably could, therefore, provide an approach to disrupting NF-�B sig-naling and, thus, potentially nullify oncogenic activity.

Acknowledgments—We thank K. S. Doctor and P. Godoi for assistance withhomology modeling, Z. Ronai, C.Ware, and K.White for provision of reagents,M. Hanaii and J. Valois for assistance with manuscript preparation, andG. Salvesen and B. Eckelman for critical reading of the manuscript.

REFERENCES1. Salvesen, G. S., and Duckett, C. S. (2002) Nat. Rev. Mol. Cell Biol. 3, 401–4102. Deveraux, Q. L., and Reed, J. C. (1999) Genes Dev. 13, 239–2523. Shu, H. B., Takeuchi, M., and Goeddel, D. V. (1996) Proc. Natl. Acad. Sci. U. S. A. 93,

13973–139784. Rothe, M., Pan, M. G., Henzel, W. J., Ayres, T. M., and Goeddel, D. V. (1995) Cell 83,

1243–12525. Li, X., Yang, Y., and Ashwell, J. D. (2002) Nature 416, 345–3476. Roy, N., Deveraux, Q. L., Takahashi, R., Salvesen, G. S., and Reed, J. C. (1997) EMBO

J. 16, 6914–69257. Hu, S., and Yang, X. (2003) J. Biol. Chem. 278, 10055–100608. Srinivasula, S. M., Hegde, R., Saleh, A., Datta, P., Shiozaki, E., Chai, J., Lee, R. A.,

Robbins, P. D., Fernandes-Alnemri, T., Shi, Y., and Alnemri, E. S. (2001)Nature 410,112–116

9. Deveraux, Q. L., Leo, E., Stennicke, H. R., Welsh, K., Salvesen, G. S., and Reed, J. C.(1999) EMBO J. 18, 5242–5251

10. Deveraux, Q. L., Roy, N., Stennicke, H. R., Van Arsdale, T., Zhou, Q., Srinivasula,S. M., Alnemri, E. S., Salvesen, G. S., and Reed, J. C. (1998) EMBO J. 17, 2215–2223

11. Schimmer, A. D., Welsh, K., Pinilla, C., Wang, Z., Krajewska, M., Bonneau, M. J.,Pedersen, I. M., Kitada, S., Scott, F. L., Bailly-Maitre, B., Glinsky, G., Scudiero, D.,Sausville, E., Salvesen, G., Nefzi, A., Ostresh, J. M., Houghten, R. A., and Reed, J. C.(2004) Cancer Cells 5, 25–35

12. Kipp, R. A., Case, M. A., Wist, A. D., Cresson, C. M., Carrell, M., Griner, E., Wiita, A.,Albiniak, P. A., Chai, J., Shi, Y., Semmelhack, M. F., and McLendon, G. L. (2002)Biochemistry 41, 7344–7349

13. Schwede, T., Kopp, J., Guex, N., and Peitsch, M. C. (2003) Nucleic Acids Res. 31,3381–3385

14. Guex, N., and Peitsch, M. C. (1997) Electrophoresis 18, 2714–272315. Peitsch, M. C., Wells, T. N., Stampf, D. R., and Sussman, J. L. (1995) Trends Biochem.

Sci. 20, 82–8416. Samuel, T., Okada, K., Hyer,M.,Welsh, K., Zapata, J.M., andReed, J. C. (2005)Cancer

Res. 65, 210–21817. Liu, Z., Sun, C., Olejniczak, E. T., Meadows, R. P., Betz, S. F., Oost, T., Herrmann, J.,

E3 Ligase Functions of cIAP1 for TRAF2 and SMAC

JANUARY 13, 2006 • VOLUME 281 • NUMBER 2 JOURNAL OF BIOLOGICAL CHEMISTRY 1089

by guest on February 5, 2018http://w

ww

.jbc.org/D

ownloaded from

Wu, J. C., and Fesik, S. W. (2000) Nature 408, 1004–100818. Huang, Y., Rich, R. L., Myszka, D. G., and Wu, H. (2003) J. Biol. Chem. 278,

49517–4952219. Leo, E., Zapata, J. M., and Reed, J. C. (1999) Eur. J. Immunol. 29, 3908–391320. Li, C., Norris, P. S., Ni, C. -Z., Havert, M. L., Chiong, E. M., Tran, B. R., Cabezas, E.,

Cheng, G., Reed, J. C., Satterthwait, A. C., Ware, C. F., and Ely, K. R. (2003) J. Biol.Chem. 278, 50523–50529

21. Habelhah, H., Takahashi, S., Cho, S. G., Kadoya, T., Watanabe, T., and Ronai, Z.(2004) EMBO J. 23, 322–332

22. Rothe,M., Sarma, V., Dixit, V.M., andGoeddel, D. V. (1995) Science 269, 1424–142723. Yang, Q. H., Church-Hajduk, R., Ren, J., Newton, M. L., and Du, C. (2003)Genes Dev.

17, 1487–1496

24. Park, S. M., Yoon, J. B., and Lee, T. H. (2004) FEBS Lett. 566, 151–15625. Tang, E. D., Wang, C. Y., Xiong, Y., and Guan, K. L. (2003) J. Biol. Chem. 278,

37297–3730526. Arch, R. H., Gedrich, R. W., and Thompson, C. B. (1998) Genes Dev. 12, 2821–283027. Ye, H., Park, Y., Kreishman, M., Kieff, E., and Wu, H. (1999)Mol. Cell 4, 321–33028. Wu, C. J., Conze, D. B., Li, X., Ying, S. X., Hanover, J. A., and Ashwell, J. D. (2005)

EMBO J. 24, 1886–189829. Laine, A., and Ronai, Z. (2005) Sci. STKE 2005 , Re530. Morgan, J., Yin, Y., Borowsky, A., Kuo, F., Nourmand, N., Koontz, J., Reynolds, C.,

Soreng, L., Griffin, C., Graeme-Cook, F., Harris, N., Weisenburger, D., Pinkus, G.,Fletcher, J., and Sklar, J. (1999) Cancer Res. 59, 6205–6213

31. Zhou, H., Du, M. Q., and Dixit, V. M. (2005) Cancer Cell 7, 425–431

E3 Ligase Functions of cIAP1 for TRAF2 and SMAC

1090 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 281 • NUMBER 2 • JANUARY 13, 2006

by guest on February 5, 2018http://w

ww

.jbc.org/D

ownloaded from

and John C. ReedTemesgen Samuel, Kate Welsh, Thomas Lober, Summanuna H. Togo, Juan M. Zapata

of CaspasesNecrosis Factor Receptor-associated Factor 2 and Second Mitochondrial Activator Distinct BIR Domains of cIAP1 Mediate Binding to and Ubiquitination of Tumor

doi: 10.1074/jbc.M509381200 originally published online November 10, 20052006, 281:1080-1090.J. Biol. Chem. 

  10.1074/jbc.M509381200Access the most updated version of this article at doi:

 Alerts:

  When a correction for this article is posted• 

When this article is cited• 

to choose from all of JBC's e-mail alertsClick here

Supplemental material:

  http://www.jbc.org/content/suppl/2005/11/18/M509381200.DC1

  http://www.jbc.org/content/281/2/1080.full.html#ref-list-1

This article cites 30 references, 16 of which can be accessed free at

by guest on February 5, 2018http://w

ww

.jbc.org/D

ownloaded from