The USP19 Deubiquitinase Regulates the Stability of c-IAP1 and c-IAP2*

The inhibitors of apoptosis (IAPs) are critical regulators of apoptosis and other fundamental cellular processes. Many IAPs are RING domain-containing ubiquitin E3 ligases that control the stability of their interacting proteins. However, how IAP stability is regulated remains unclear. Here we report that USP19, a deubiquitinating enzyme, interacts with cellular IAP 1 (c-IAP1) and c-IAP2. Knockdown of USP19 decreases levels of both c-IAPs, whereas overexpression of USP19 results in a marked increase in c-IAP levels. USP19 effectively removes ubiquitin from c-IAPs in vitro, but it stabilizes c-IAPs in vivo mainly through deubiquitinase-independent mechanisms. The deubiquitinase activity is involved in the stabilization of USP19 itself, which is facilitated by USP19 self-association. Functionally, knockdown of USP19 enhances TNFα-induced caspase activation and apoptosis in a c-IAP1 and 2-dependent manner. These results suggest that the self-ubiquitin ligase activity of c-IAPs is inhibited by USP19 and implicate deubiquitinating enzymes in the regulation of IAP stability.

The inhibitors of apoptosis (IAPs) 5 were initially identified as baculovirus-encoded products that prevent apoptosis of infected insect cells (1,2). The cellular homologues of baculoviral IAPs have since been found in a wide range of organisms and are involved in both apoptotic and non-apoptotic processes (3)(4)(5)(6). The function of IAPs is largely governed by two zinc-binding domains: the baculoviral IAP repeat (BIR) and the really interesting new gene (RING) domains. Each IAP has at least one copy, and often three tandem copies, of the BIR domain, the signature motif of this family. BIR domains and the flanking sequences mediate the interaction of IAPs with their target proteins. The RING domain, which is present in about half of all mammalian IAPs, endows these IAPs with the ubiquitin E3 ligase activity and determines the consequence of these interactions (4,(7)(8)(9). For apoptosis regulation, the BIR domains and the flanking regions of some IAPs, particularly the mammalian X-linked IAP (XIAP) and the Drosophila IAP1, directly bind to and inhibit caspases (10), the apoptotic proteases that orchestrate the initiation and execution of apoptosis (11,12). The E3 ligase activity of XIAP and Drosophila IAP1, which mediates proteasomal degradation of caspases, is further needed for effective suppression of apoptosis (13,14).
The functions of IAPs are nevertheless diverse. In this context, the most extensively investigated IAPs are perhaps the mammalian cellular IAP 1 (c-IAP1) and c-IAP2. c-IAPs share high sequence similarity and were initially identified as part of the intracellular signaling complex that is formed following activation of the tumor necrosis factor receptor (15). Unlike XIAP, c-IAP1 and c-IAP2 exhibit minimal binding to caspases and may not play an important role in the inhibition of these proteases. c-IAPs do interact with a multitude of overlapping and distinct substrates, including those involved in apoptosis, NF-B signaling, and oncogenesis, and mediate the ubiquitination of these substrates through their robust ubiquitin E3 ligase activity (7)(8)(9). c-IAPs ubiquitinate Smac/ DIABLO, a mitochondrial protein released to the cytoplasm on apoptosis induction (16,17), and target it for proteasomal degradation leading to apoptosis inhibition (9). In the canonical NF-B pathway activated by TNF␣, c-IAP1 and c-IAP2 promote K63 poly-ubiquitination of RIP1, which enhances NF-B activation (18,19). By contrast, in the noncanonical NF-B pathway, c-IAPs are responsible for the ubiquitination and degradation of NF-B-inducing kinase with the consequential attenuation of NF-B activation (20 -23). The importance of c-IAPs in human disease is highlighted by the involvement of c-IAP2 in mucosa-associated lymphoid tissue (MALT) lymphoma, the most common type of human lymphoma that arises in extranodal sites (24). A high percentage of MALT lymphomas express a c-IAP2-MALT1 fusion protein generated by the t(11;18) chromosomal translocation. The target of c-IAP2 related to MALT lymphomas is likely Bcl10, an essential component for antigenic signaling to NF-B (25,26). Regulation of Bcl10 by c-IAP2 is disrupted in cells with the c-IAP2-MALT fusion protein (25,26), which may contribute to hyperactive NF-B and uncontrolled B-cell proliferation. Although this func-tion implied a role for c-IAP2 in tumor suppression, the importance of c-IAP1 in tumorigenesis is suggested by its overexpression in certain human tumors and by its ability to target the c-Myc inhibitor Mad1 for proteasomal degradation (27).
c-IAPs also regulate their own levels or each other's through their ubiquitin E3 activity. c-IAPs, as well as XIAP, undergo self-ubiquitination and proteasomal degradation during glucocorticoid-induced T-cell death (7). Binding to Smac can also induce c-IAP self-ubiquitination and degradation (28). In addition, c-IAP1 promotes the degradation of c-IAP2 via ubiquitination (29,30). Of note, small compound Smac mimetics prompt the degradation of c-IAP1 and c-IAP2, causing NF-Bmediated TNF␣ production and subsequent TNF␣-dependent apoptosis in certain tumor cells (20,21,31,32). Smac mimetics are currently in preclinical and clinical tests for cancer therapy (6,33).
Despite increasing knowledge on c-IAP-mediated degradation of various target proteins, how self-ubiquitination of c-IAPs is normally prevented remains poorly understood. Protein ubiquitination is a reversible process, and increasing evidence indicates the importance of deubiquitinating enzymes (DUBs) in ubiquitin-dependent pathways (34,35). Two major classes of DUBs have been identified, the ubiquitin C-terminal hydrolases and ubiquitin-specific processing proteases (UBPs), both of which are cysteine proteases. Although ubiquitin C-terminal hydrolases are relatively small in size and release ubiquitin from its precursors and small adducts, UBPs are large proteins and remove ubiquitin from poly-ubiquitinated proteins. One of the UBPs, USP19, was previously shown to regulate the stability of the cell cycle regulator p27 Kip and to rescue the substrates of endoplasmic reticulum-associated degradation (36,37). In this study, we identify USP19 as a c-IAP1 and 2-interacting protein. USP19 prevents ubiquitination of c-IAP1 and c-IAP2 and stabilizes them. Our finding reveals a role for DUBs in controlling the stability of the c-IAP ubiquitin E3 ligases.
Affinity Purification-The FLAG-tagged c-IAP2 BIR2 ϩ 3 (amino acids 101-346) was expressed in HEK293T cells through transient transfection and purified to apparent homogeneity via anti-FLAG mAb M2 conjugated to agarose beads. The immobilized FLAG-c-IAP2 BIR protein was then incubated with the S100 fraction of HeLa S3 cells as described (38,39). As a control, the HeLa S100 fraction was also incubated with the anti-FLAG beads. After extensive washing, the bound proteins were eluted, and proteins that specifically associated with FLAG-c-IAP2 BIR2 ϩ 3 were analyzed by mass spectrometry at the Proteomic Core Facility of the Abramson Cancer Center at the University of Pennsylvania.
Sequence Analysis and Northern Blot-In the USP19 plasmid KIAA0891, an in-frame stop codon before the first ATG is not found in the sequence of USP19 but exists in a human EST clone containing additional nucleotides at the 5Ј end (No. CB044698). Thus, the sequence in KIAA0891 is likely to be the full-length USP19 sequence. The entire open reading frame was sequenced. The protein domains of USP19 were analyzed by using the SMART program (available on-line). For Northern analysis, human 12-well multiple tissue (Clontech, catalogue no. 636818) and cancer cell line (Clontech, catalogue no. 636804) blots were hybridized with a [ 32 P]dCTP-labeled cDNA probe spanning the C-terminal segment (nucleotides 3140 -3960) of the human USP19 open reading frame according to the manufacturer's instruction.
RNA Interference-USP19 shRNAs and control shRNA were purchased from Open Biosystems. To generate lentiviruses expressing USP19 and control shRNAs, HEK293T cells grown on a 6-cm dish were transfected with 2 g of PLKO.1 USP19 shRNAs or control vector, 2 g of pREV, 2 g of pGag/Pol/PRE, and 1 g of pVSVG. 24 h after transfection, cells were cultured with DMEM medium containing 20% FBS for an additional 24 h. The culture medium containing lentivirus particles was centrifuged at 1000 ϫ g for 5 min, and viruses in the supernatant were used for infection.
Immunoprecipitation and Western Blot-HEK293T cells were transfected with the indicated plasmids. 24 h after transfection, cells were harvested and lysed in the IP lysis buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1.5 mM MgCl 2 , 20 M MG132, 10% glycerol, 1% Triton X-100, and protease inhibitors). The soluble fractions were incubated with anti-FLAG M2 or anti-HA beads for 2 h at 4°C. The immunoprecipitates were then washed five times and resolved by SDS-PAGE followed by Western blot analysis using rabbit anti-FLAG or anti-HA antibody. To detect the interaction of endogenous USP19 and c-IAP2, Jurkat cells were treated with MG132 for 4 h. Cell lysates were incubated overnight at 4°C with anti-USP19 antibody or control rabbit IgG antibody bound to protein A-agarose beads. The immunoprecipitated proteins were analyzed by Western blot with anti-USP19 and anti-c-IAP2 antibodies.
Immunofluorescence-HeLa cells grown on glass coverslips were transfected with indicated plasmids. 24 h after transfection, cells were washed once with PBS, fixed with 4% paraformaldehyde, and permeabilized with 0.01% Triton X-100 (in PBS). After being blocked with 1% BSA (in PBS) for 1 h, cells were incubated at room temperature with anti-FLAG or anti-HA antibody for 1 h and with rhodamine-conjugated antimouse or FITC-conjugated anti-rabbit IgG secondary antibodies for an additional 1 h. The cells were washed twice with PBS containing 0.1% Tween 20, dried in 100% EtOH, and mounted with Vectashield containing DAPI. The cells were visualized by fluorescence microscopy (Leica).
In Vivo and in Vitro Deubiquitination Assays-HEK293T cells expressing FLAG-c-IAP2, FLAG-c-IAP2 and HA-USP19, or FLAG-c-IAP2 and HA-USP19 CA were lysed in 1% SDS (in 50 mM Tris-HCl, pH 7.6, and 120 mM NaCl) and boiled for 5 min at 100°C. The lysates were diluted 1:10 in IP-lysis buffer and immunoprecipitated by anti-FLAG M2 beads as described above. The immunoprecipitated proteins were analyzed by Western blot with polyclonal anti-FLAG antibody. For in vitro deubiquitination experiments, FLAG-USP19 and FLAG-USP19 CA were expressed in HEK293T cells and purified as described (38,39). The purified proteins were eluted with 150 ng/ml 3ϫFLAG peptide. The proteins were then incubated with tetra-ubiquitin (0.75 g) in deubiquitination buffer (50 mM Tris-HCl, pH 7.4, 1 mM MgCl 2 , and 1 mM DTT) at 37°C for the indicated time. USP19 was also incubated in the presence of 5 mM iodoacetate. The samples were resolved on a 4 -20% gradient gel and analyzed by Coomassie Blue staining. For c-IAP2 deubiquitination, modified c-IAP2 was purified from HEK293T cells that had been transfected for 24 h and treated with 20 M ALLN for 4 h. After being eluted from M2 beads, c-IAP2 was incubated with USP19 and USP19 CA.
Apoptosis Assay-MDA-MB 231 cells expressing control shRNA or USP19 specific shRNA were treated with or without TNF␣ (20 ng/ml) for 20 h. Cells were stained with Hoechst 33342, and apoptosis was determined by counting the apoptotic cells with the aberrant nuclei staining.
Reproducibility-All data, with the exception of the initial affinity purification of IAP2-interacting proteins and the Northern blot analysis, are representatives of two to three independent experiments.

RESULTS
USP19 Interacts with c-IAPs and XIAP-To better understand the regulation of c-IAP stability, we sought to identify c-IAP2-interacting proteins using an affinity purification approach (38,39). Because the BIR region is the main proteininteraction domain of IAPs, we expressed a FLAG-tagged c-IAP2 protein containing the second and third BIR domains (FLAG-c-IAP2 BIR2 ϩ 3) in HEK293T cells and purified it to apparent homogeneity using anti-FLAG antibody. The FLAG-c-IAP2 BIR2 ϩ 3 protein immobilized on beads was then incubated with S100 extract of HeLa S3 cells. Proteins that specifically bound to the c-IAP2 BIR2 ϩ 3 beads, but not the control beads, were analyzed by mass spectrometry. One protein was found to be USP19. Like the other UBPs, USP19 contains the USP enzymatic domain with a conserved catalytic cysteine-histidine dyad. In addition, USP19 contains a transmembrane domain at the C terminus (supplemental Fig. S1) (37,40).
To examine the expression profile of USP19 mRNA, a human multiple tissue Northern blot was hybridized with a radioiso-tope-labeled USP19 cDNA probe. A single transcript ϳ5 kb in size was detected in virtually all tissues examined, with high expression found in the heart, skeletal muscle, brain, kidney, and placenta; medium expression in the spleen, liver, small intestine, thymus, and leukocytes; and low expression in the colon and lung (supplemental Fig. S2). In addition, the USP19 transcript was detected in various human tumor cell lines with comparable levels of expression, except in lung carcinoma A539 and promyelocytic leukemia HL60 cells where less expression was observed (supplemental Fig. S2).
To confirm the interaction of c-IAP2 with USP19, FLAGtagged c-IAP2 was expressed together with HA-tagged USP19 in HEK293T cells. A co-immunoprecipitation assay using anti-HA antibody showed a specific interaction between these two proteins ( Fig. 1A and supplemental Fig. S3A). The interaction was verified in a reciprocal co-immunoprecipitation experiment using anti-FLAG antibody (Fig. 1B). We next examined the interaction between endogenous c-IAP2 and USP19. When lysates from Jurkat cells were incubated separately with anti-USP19 antibody and a control antibody, c-IAP2 was detected in the anti-USP19, but not the control, immunoprecipitates (Fig. 1C). The specificity of the anti-c-IAP2 antibody was underscored by the observation that, when cells were treated with the Smac mimetics BV6, the detected c-IAP2 signal rapidly disappeared as expected (20) (supplemental Fig.  S3B). Taken together, these results indicate that c-IAP2 interacts with USP19. Given the structural similarity between c-IAP2 and c-IAP1, we tested whether c-IAP1 binds to USP19. As shown in Fig. 1A, c-IAP1 also interacted with USP19, but not as strongly as c-IAP2 (Fig. 1A). In addition, XIAP associated with USP19 (Fig. 1B).
We next examined whether USP19 co-localizes with c-IAP2. When expressed individually, c-IAP2 mainly localized in the nucleus, whereas USP19 was exclusively associated with cytoplasmic structures resembling endoplasmic reticulum membrane, consistent with previously reports (37,41). In the presence of USP19, c-IAP2 was re-localized to the cytoplasm, exhibiting a localization pattern similar to USP19 (Fig. 1D). This result indicates the interaction of USP19 and c-IAP2 in cells.
Structural Determinants of the c-IAP2-USP19 Interaction-To delineate the structural determinants of the c-IAP2-USP19 interaction, we used a panel of c-IAP2 and USP19 mutants ( Fig.  2A). The c-IAP2 N-terminal region containing all three BIR domains (N) showed a strong interaction with USP19, whereas the C-terminal region containing the CARD and RING domains (C) showed no interaction (Fig. 2B). Furthermore, deletion of either the BIR1 (⌬N100 and BIR2 ϩ 3) or the BIR3 (BIR1 ϩ 2) domain didn't affect the c-IAP2-USP19 interaction (Fig. 2C). Therefore, the c-IAP2 BIR2 domain likely mediates the interaction of c-IAP2 with USP19.
USP19 contains a ubiquitin-specific protease domain in its C terminus that includes a catalytic Cys residue at amino acid 506 ( Fig. 2A). Mutation of this Cys to Ala (CA) did not impair the interaction of USP19 with c-IAP2 N (Figs. 1B and 2D). Furthermore, both the region N-terminal to the USP domain (N) and the region containing the USP and the transmembrane domain (C) were able to associate with c-IAP2 N (Fig. 2D). For these experiments, c-IAP2 N was used due to a more stable expression level compared with full-length c-IAP2. These data suggest that both N-and C-terminal regions of USP19 interact with c-IAP2.

USP19 Deubiquitinates c-IAP2 in Vitro and Prevents c-IAP2 Ubiquitination in Vivo through Both DUB-dependent and -independent
Mechanisms-To confirm the deubiquitinase activity of USP19, we purified FLAG-USP19 to apparent homogeneity and incubated it with a tetra-ubiquitin chain. As shown in Fig. 3A, USP19, but not USP19 CA, cleaved the tetra-ubiquitin chain to mono-ubiquitin. The activity of USP19 was blocked by iodoacetate, a cysteine-specific alkylating agent that inhibits deubiquitinating enzymes.
To determine whether USP19 can remove ubiquitin from conjugated c-IAP2, poly-ubiquitinated c-IAP2 was generated by transiently overexpressing FLAG-c-IAP2 in HEK293T cells in the presence of the proteasome inhibitor ALLN and subsequently affinity-purified. The purified c-IAP2 protein was almost completely deubiquitinated by the purified USP19 protein (Fig. 3B). Again, this activity of USP19 was abolished by iodoacetate, and USP19 CA showed no DUB activity on ubiquitinated c-IAP2.
We next investigated the effect of USP19 on c-IAP2 ubiquitination in vivo. c-IAP2 was strongly self-ubiquitinated upon overexpression in HEK293T cells. However, when coexpressed with USP19, c-IAP2 ubiquitination was virtually abolished (Fig. 3C). Interestingly, USP19 CA also inhibited c-IAP2 ubiquitination, although to a lesser extent compared with USP19. Because USP19 CA lacked any detectable de-ubiquitinating enzyme activity (Fig. 3, A and B), it likely reduces c-IAP2 ubiquitination through a DUB-independent mechanism.
USP19 Stabilizes c-IAP1 and c-IAP2-To examine the effect of USP19 on the expression of endogenous c-IAP1 and c-IAP2, we knocked down USP19 in HeLa and MDA-MB231 cells using small hairpin (sh) RNA. This led to a significant decrease in the levels of endogenous c-IAP2 and, to a lesser extent, the level of endogenous c-IAP1 (Fig. 4A). Knockdown of USP19 also reduced endogenous c-IAP2 levels in Jurkat cells (supplemental Fig. S3C). Next we increased the amount of USP19 through transient transfection. Upon overexpression, USP19 markedly increased the levels of c-IAP2 (Fig. 4B). Notably, the DUB-deficient USP19 CA similarly increased c-IAP2 levels. To determine whether USP19 prolonged the half-life of c-IAP proteins, FLAG-c-IAP1 and FLAG-c-IAP2 were expressed alone and together with HA-USP19 or HA-USP19 CA, and cells were treated with the protein synthesis inhibitor cycloheximide for variable lengths of time. c-IAP1 and c-IAP2 were strongly stabilized by both USP19 and USP19 CA (Fig. 4C). In contrast, USP19 and USP19 CA did not influence the stability of the tumor suppressor p53 (Fig. 4C), which is a labile protein, underscoring the specific effect of USP19 and USP19 CA on c-IAPs. Conversely, knockdown of USP19 by shRNA noticeably shortened the half-life of both endogenous c-IAP1 and c-IAP2 (Fig.  4D). Taken together, these results show that USP19 controls the stability of c-IAP1 and c-IAP2. Moreover, the effect of USP19 overexpression on c-IAP stability is largely independent of its DUB activity. Cell lysates were subjected to immunoprecipitation with anti-FLAG M2 beads followed by Western blot analysis. C, interaction of endogenous USP19 and c-IAP2. Jurkat cell lysates were subjected to immunoprecipitation with anti-USP19 antibody. Immunoprecipitates were analyzed by Western blot using anti-c-IAP2 antibody. *, a nonspecific band. D, co-localization of exogenous USP19 and c-IAP2. HeLa cells expressing FLAG-c-IAP2 and HA-USP19 individually or in combination were immunostained with anti-FLAG and/or anti-HA antibody followed by Rhodamine-conjugated anti-mouse and/or FITC-conjugated anti-rabbit IgG.

USP19 Associates with Itself and Is Stabilized by Its Own
Deubiquitinase Activity-Given that USP19 does not seem to rely on its DUB activity to stabilize c-IAPs (Fig. 4, B and C), we asked whether the DUB activity of USP19 plays a role in regulating its own ubiquitination. When expressed together with HA-ubiquitin in HEK293T cells, USP19 CA was readily ubiquitinated (Fig. 5A). This ubiquitination was further enhanced upon treatment with the proteasome inhibitor ALLN. In contrast, no ubiquitination was detected on USP19 even in the presence of ALLN. Therefore, USP19 likely removes associated ubiquitin moieties via self-deubiquitination. Correlating with their ubiquitination status, USP19 was noticeably more stable than USP19 CA (Fig. 5B). To investigate how USP19 might be stabilized by its de-ubiquitinase activity, we tested whether USP19 can self-interact. FLAG-USP19 was expressed with HA-USP19 or HA-USP19 CA in HEK293T cells. An immunoprecipitation assay showed that FLAG-USP19 could associate with both HA-USP19 and HA-USP19 CA (Fig. 5C). In contrast, USP19 did not interact with another DUB, USP7/HAUSP (supplemental Fig. S3D). As expected from the ability of USP19 to selfinteract, USP19 increased the levels of USP19 CA in a dosedependent manner when they were co-expressed in cells (Fig. 5D). Collectively, these results suggested that USP19 stabilizes itself through self-association and intermolecular deubiquitination.
USP19 Inhibits TNF␣-induced Apoptosis via the Stabilization of c-IAPs-c-IAPs inhibit TNF␣-mediated apoptosis, a property underlying the cytotoxic effect of Smac mimetics in cancer cells (20,21,31,32). To examine the physiological significance of USP19-mediated stabilization of c-IAPs, USP19 was knocked down in MDA-MB231 breast cancer cells. Compared with control cells, USP19 knockdown cells showed elevated levels of apoptosis (from 10% to 21%) in response to treatment with TNF␣ (Fig. 6A, lane 4 versus 2). Accordingly, USP19 knockdown cells also exhibited higher levels of caspase-8 activation and poly(ADP-ribose) polymerase cleavage compared with control cells. TNF␣ treatment effectively increased the levels of c-IAPs, particularly c-IAP2. This increase was likely due to TNF␣-mediated activation of NF-B, which controls the expression of c-IAPs (42). Nevertheless, the increase in IAP expression was less in USP19 knockdown cells compared with control cells (Fig.  6A). To further test whether USP19 may have an anti-apoptotic function through stabilizing c-IAP1 and c-IAP2, FLAG-c-IAP1 and FLAG-c-IAP2 were individually transfected into the USP19 knockdown and control cells. Both c-IAP1 and c-IAP2 prevented the increase in TNF␣-induced apoptosis caused by USP19 knockdown (Fig. 6B, lanes 12 and  10 versus 8). c-IAP1 and c-IAP2 also decreased TNF␣-induced poly(ADP-ribose) polymerase cleavage and caspase-8 activation in USP19 knockdown cells. Compared with  c-IAP2, c-IAP1 was less potent in reversing apoptosis that was caused by USP19 knockdown, consistent with their relative binding strengths to USP19 (Fig. 1A). Taken together, these data suggest that USP19-mediated stabilization of c-IAPs inhibits TNF␣-induced apoptosis.

DISCUSSION
c-IAP1 and 2 are ubiquitin ligases that regulate the stability of a variety of apoptotic and non-apoptotic proteins. An intriguing issue remains on how ubiquitination and proteasomal degradation of c-IAPs are normally prevented. The current study identifies USP19 as a stabilizer for c-IAPs with the consequential inhibition of apoptosis, revealing a role for a deubiquitinating enzyme in the regulation of IAP stability. Stabilization of ubiquitin E3 ligases by deubiquitinating enzymes has been suggested by previous studies. For example, the stability of Mdm2, an oncogenic E3 ligase and the major antagonist of the tumor suppressor p53, is controlled by the deubiquitinating enzyme HAUSP via the adaptor protein Daxx (39). USP19-mediated stabilization of c-IAPs further underscores the dynamic interplay between ubiquitin ligases and deubiquitinating enzymes.
Notably, USP19 appears to stabilize c-IAPs through a protease-independent mechanism, at least upon overexpression (Figs. 3 and 4). The precise nature remains to be determined, but conceivably, USP19 might mask Lys residues on c-IAPs that are critical for ubiquitination or prevent the binding of a ubiquitin conjugating enzyme (E2) to c-IAPs. Regardless, this deubiquitinase-independent mechanism may be important considering that c-IAPs, like other E3 ligases, could in theory constantly self-ubiquitinate. By blocking the self-addition of ubiquitin, USP19 would effectively prevent a futile cycle of ubiquitination and deubiquitination. The deubiquitinating activity might then reverse any small amount of ubiquitination, functioning as a fail-safe mechanism to prevent aberrant activation of apoptosis. It will be interesting to determine whether other deubiquitinating enzymes also function in a similar manner to prevent the self-ubiquitination of E3 ligases. The deubiquitinase activity of USP19 does appear to control its own stability, which is facilitated by dimerization or oligomerization (Fig. 5), and thus may indirectly affect c-IAP levels.
Among the c-IAPs, c-IAP2 appears to be a major target of USP19. Despite their high structural homology, c-IAP1 and c-IAP2 have overlapping and distinct targets and functions. For example, c-IAP1, but not c-IAP2, mediates the ubiquitination of TRAF2 and Mad1 (8,27), whereas only c-IAP2 is involved in MALT lymphomas (25,26). The preferential binding of USP19 to c-IAP2 suggests that the stability of c-IAPs may be differen-tially regulated. Although XIAP can also interact with USP19, the consequence of this interaction is currently unclear because XIAP is a much weaker E3 ligase compared with c-IAP1 and c-IAP2 (9) and may rely less on a deubiquitinating enzyme for stabilization. However, it remains possible that, by binding to XIAP (and c-IAPs), USP19 might influence the ubiquitination status of the IAP targets.
Currently, IAP antagonists, such as compounds mimicking the IAP antagonist Smac/DIABLO, are being extensively investigated as a treatment for cancer. Although Smac was thought to mainly inhibit XIAP, Smac mimetics induce c-IAP degradation and generate autocrine TNF␣ apoptotic signaling (20,21,31,32). However, the efficacy of these mimetics varies among different tumors (31), and the cause of this variation is a highly important, yet poorly understood, issue. It will be interesting to determine whether USP19 influences the effectiveness of Smac mimetics in cancer therapy.