USP11-dependent selective cIAP2 deubiquitylation and stabilization determine sensitivity to Smac mimetics

Abstract

Given their crucial role in apoptosis suppression, inhibitor of apoptosis proteins (IAPs) have recently become attractive targets for cancer therapy. Here, we report that cellular IAP2 (cIAP2) is specifically stabilized in several cancer cell lines, leading to resistance to Smac mimetics, such as BV6 and birinapant. In particular, our results showed that cIAP2 depletion, but not cIAP1 depletion, sensitized cancer cells to Smac mimetic-induced apoptosis. Ubiquitin-specific protease 11 (USP11) is a deubiquitylase that directly stabilizes cIAP2. USP11 overexpression is frequently found in colorectal cancer and melanoma and is correlated with poor survival. In our study, cancer cell lines expressing high levels of USP11 exhibited strong resistance to Smac mimetic-induced cIAP2 degradation. Furthermore, USP11 downregulation sensitized these cells to apoptosis induced by TRAIL and BV6 and suppressed tumor growth in a xenograft model. Finally, the TNFα/JNK pathway induced USP11 expression and maintained cIAP2 stability, suggesting an alternative TNFα-dependent cell survival pathway. Collectively, our data suggest that USP11-stabilized cIAP2 may serve as a barrier against IAP-targeted clinical approaches.

Apoptosis is an inherent cell death program that is crucial for various physiological processes such as development, the immune response, and tumorigenesis.¹ This process is finely tuned by numerous cellular signaling pathways involving hundreds of pro-apoptotic and anti-apoptotic factors.^{2, 3} Inhibitor of apoptosis proteins (IAPs) are a conserved protein family containing the baculoviral IAP repeat (BIR) domain.⁴ There are eight human IAP proteins, including cellular IAP1 (cIAP1/BIRC2), cIAP2/BIRC3, X chromosome-linked IAP (XIAP/BIRC4), and melanoma IAP (ML-IAP/BIRC7).⁵ IAPs such as XIAP can exert their anti-apoptotic function through the BIR domain, which directly interacts with caspases.⁵ In addition, several IAPs contain a RING domain with E3 ubiquitin ligase activities, which are crucial for apoptosis suppression. In particular, the E3 ligase activities of cIAP1/2 are necessary to regulate tumor necrosis factor receptor (TNFR) signaling.⁶ Upon TNFR activation, cIAP1/2 is recruited to TNFR through TNFα receptor-associated factor 2 (TRAF2), leading to K63-linked polyubiquitylation of receptor interacting protein kinase 1 (RIPK1), which is essential for NF-κB-mediated cell survival.⁷ The lack of RIPK1 polyubiquitylation via cIAP1/2 depletion or the presence of CYLD deubiquitylase triggers TNFR complex IIa formation, thereby inducing caspase-8-dependent apoptosis.⁸ In addition, cIAP1/2 prevents the formation of the RIPK1-containing death complex ripoptosome in response to several stimuli including CD95, TNFα-related apoptosis-inducing ligand (TRAIL), genotoxic stress, and Toll-like receptor (TLR) activation.^{9, 10, 11, 12, 13}

IAPs are frequently overexpressed in various human cancers, and their expression is associated with chemoresistance and poor clinical outcome.⁶ Therefore, inhibiting IAP function is an attractive strategy to treat cancer through the induction of apoptosis.^{5, 14} Upon apoptotic stimuli, IAPs are inhibited by the second mitochondria-derived activator of caspases (Smac),⁵ and this discovery led to the development of Smac mimetic peptides using the IAP binding motif containing four amino acids (Ala-Val-Pro-Ile). These peptides were shown to sensitize cells to apoptotic stimuli and efficiently suppress tumor growth in a xenograft model.^{15, 16} Subsequently, a number of small-molecule compounds mimicking the Smac mimetic peptide (Smac mimetics) were developed with improved pharmacological properties and IAP-binding affinity. Interestingly, Smac mimetics, such as BV6 and compound A, were found to induce autoubiquitylation and degradation of cIAP1/2.^{17, 18} Furthermore, cIAP1/2 depletion with Smac mimetics activates the non-canonical NF-κB pathway to induce autocrine TNFα production, which is essential for Smac mimetic-induced apoptotic cell death.^{18, 19}

Because cIAP1 and cIAP2 show functional redundancy in TNFα-mediated survival, the depletion of both proteins is usually required for effective induction of cell death upon TNFα treatment.^{20, 21} However, there are several reports showing that cIAP2 expression, but not cIAP1 expression, renders cells resistant to Smac mimetic-induced cell death.^{20, 21} For example, cIAP2 upregulation via phosphoinositide 3-kinase (PI3K) upon compound 3 treatment in certain cell lines was shown to facilitate apoptosis evasion.²² In addition, treatment with compounds A and C led to cIAP1 dimerization, without cIAP2 dimerization, resulting in the autoubiquitylation and subsequent degradation of cIAP1. These findings may explain why cIAP1 is degraded more efficiently than cIAP2 upon treatment with Smac mimetics.²³ Alternatively, because cIAP2 degradation requires cIAP1, cIAP2 may become more stable when cIAP1 is depleted using Smac mimetics.²⁴ Direct cIAP deubiquitylation by OTUB1 or USP19 has been suggested to be responsible for cIAP stabilization;^{25, 26} however, these previous studies did not focus on the difference in stabilization between cIAP1 and cIAP2 and only provided general deubiquitylation-dependent mechanisms.^{25, 26}

While several studies have supported hypotheses for how cIAP2 survives in the presence of Smac mimetics, numerous independent studies have also shown that cIAP2 can be efficiently degraded by Smac mimetics in various cell lines.^{27, 28, 29, 30, 31, 32} These observations suggest the existence of other factors that specifically regulate cIAP2 stability upon Smac mimetic treatment. In this study, we propose a new mechanism involving USP11-mediated cIAP2 regulation. We found that the differential destabilization of cIAP1 and cIAP2 is dependent on the presence of the cIAP2-specific deubiquitylase USP11. Mechanistically, USP11 can protect cIAP2 from Smac mimetic-mediated degradation, rendering cell lines with high USP11 expression unresponsive to Smac mimetic treatment. However, USP11 downregulation sensitized these cells to TNFα- or TRAIL-induced apoptosis in the presence of Smac mimetic and further suppressed tumor growth in a xenograft model. Corroborating these data, USP11 overexpression was observed in colon cancer and melanoma patients with poor clinical outcome. Finally, the TNFα/c-Jun N-terminal kinase (JNK) pathway induced USP11 expression, which was necessary for cIAP2 protein stabilization and its anti-apoptotic function. Thus, the identification of cIAP2-specific deubiquitylation indicates that more elaborate strategies should be developed for pharmaceutical therapies targeting cIAPs.

Results

TNFα stimulation induces cIAP2 expression and renders cells resistant to Smac mimetic-induced apoptotic cell death

Smac mimetics such as BV6 induce IAP autoubiquitylation and degradation, thereby sensitizing cancer cells to apoptotic stimuli.^{17, 18} Consistently, we observed that cIAP1 and cIAP2 showed similar sensitivities to BV6 in TNFα-treated HT-29 cells (Figures 1a and b). Interestingly, BV6-mediated cIAP2 degradation was extensively protected in H1299 cells compared with cIAP1 degradation in the presence or absence of TNFα treatment (Figures 1a–c). In contrast to cIAP2, cIAP1 expression was not affected by TNFα stimulation and showed exhaustive destabilization after BV6 co-treatment (Figure 1c). It has previously been reported that several Smac mimetics can induce cIAP2 expression in certain conditions.^{22, 33} To test whether BV6 increased cIAP2 expression in our conditions, we stimulated cells with TNFα in the presence or absence of BV6. cIAP2 induction following TNFα treatment was not affected by BV6 treatment, excluding the possibility of BV6 affecting cIAP2 transcriptional levels (Figure 1d). Finally, unlike cIAP1, cIAP2 induced by TNFα pretreatment in H1299 cells was resistant to BV6 and cycloheximide (CHX) treatment, suggesting that cIAP2 was indeed protected from protein degradation (Figure 1e). H1299 cells stimulated with TNFα and BV6 exhibited slight activation of caspase-8 and -3, possibly due to the comparable expression and stabilization of cIAP2 (Figure 1f). Furthermore, cIAP2 knockdown, but not cIAP1 knockdown, facilitated caspase activation and subsequent TNFα/BV6-induced cell death (Figures 1f and g). We further tested an additional Smac mimetic, birinapant, which was recently reported to induce efficient degradation of both cIAP1 and cIAP2.²⁸ Similar to BV6, birinapant also exhibited limited cIAP2 degradation in H1299 cells, and cIAP2 depletion, but not cIAP1 depletion, decreased H1299 cell viability upon TNFα and birinapant treatment (Supplementary Figures S1a and b). Overall, these observations suggest that cIAP2 induction and stabilization by TNFα may inhibit Smac mimetic-induced apoptosis in certain cancer cell lines.

Figure 1.

Inefficient cIAP2 degradation upon BV6 treatment prevents caspase-dependent apoptosis. (a) HT-29 cells were pretreated with 5 ng/ml TNFα for 12 h and treated with increasing concentrations of BV6 for 30 min. H1299 cells were treated with BV6 in the absence of TNFα as indicated. cIAP1 and cIAP2 protein levels were determined by WB. (b) H1299 and TNFα-treated HT-29 cells were treated with 1 μM BV6 for the indicated time periods, and the protein levels were determined by WB. (c and d) H1299 cells were treated with 20 ng/ml TNFα and 1 μM BV6, as indicated. cIAP1/2 protein levels were determined by WB (c), and cIAP1/2 mRNA levels were determined by quantitative reverse-transcriptase PCR (qRT-PCR) (d) (data represent the mean±S.D.; n=3, n.s.=non-significant). (e) H1299 cells were pretreated with 20 ng/ml TNFα for 6 h and treated with 1 μM BV6 in the presence of 50 μg/ml CHX. WB was performed to detect cIAP1 and cIAP2. (f) H1299 cells were transfected with 20 nM of an siRNA pool against cIAP1 or cIAP2 and treated with 20 ng/ml TNFα and 1 μM BV6 for 2 or 4 h. Caspase-8 and -3 activation was determined by WB. (g) siRNA-transfected H1299 cells were treated with TNFα and BV6 for 6 h. Cell viability was determined by measuring ATP levels using the Cell Titer Glo Luminescent Cell Viability Assay Kit (data represent the mean±S.D.; n=3, **P<0.01)

USP11, a novel cIAP2 deubiquitylating enzyme, is responsible for cIAP2 stabilization

cIAP2's resistance to Smac mimetic-induced degradation indicates that deubiquitylating enzymes (DUBs) (or deubiquitylase) may inhibit cIAP2 ubiquitylation-mediated degradation. Sixty-seven DUBs were tested in a screen for a cIAP2 deubiquitylase. Several DUBs, including USP11, USP26, USP29, USP37, OTUD1, OTUD7B, and DUB3, increased cIAP2 expression (Figure 2a; Supplementary Figure S2); however, despite their effects on cIAP2 protein levels, only USP11 and DUB3 effectively reduced cIAP2 ubiquitylation (Figures 2b and c). Among these DUBs, we focused on USP11 because it is known to regulate various signaling pathways involved in cell survival, including TNFα, TGFβ, and Notch signaling. Protein binding analyses showed that endogenous USP11 interacted with endogenous cIAP2, and their interaction was enhanced upon TNFα treatment (Figure 2d). We further demonstrated this protein interaction using ectopically expressed or recombinant proteins (Figures 2e and f).

Figure 2.

USP11 is a DUB that stabilizes cIAP2 via direct binding. (a) Relative cIAP2 protein levels normalized to 6xMyc in the presence of each DUB are shown in the graph. (b) 293FT cells, transfected with the indicated construct, were treated with 10 μM MG132 for 6 h. Cell lysates were immunoprecipitated with an anti-FLAG antibody, followed by WB. (c) 293FT cells were transfected with the indicated construct, and cIAP2 ubiquitylation was purified using Ni2⁺-NTA beads and detected by WB. (d) Lysates of H1299 cells stimulated with or without TNFα were immunoprecipitated with an anti-USP11 antibody, followed by WB. (e) 293FT cells were transfected as indicated and treated with 10 μM MG132 for 6 h. Cell lysates were immunoprecipitated with an anti-FLAG antibody, followed by WB using an anti-Myc antibody. (f) Bacterially purified GST-USP11 was incubated with His-cIAP2 and pulled down with glutathione sepharose beads. USP11-bound cIAP2 was detected by WB using an anti-cIAP2 antibody

To further analyze the roles of USP11 in cIAP2 regulation, wild-type (WT) USP11 and a catalytically inactive C318S mutant (MT) were employed. While WT and MT USP11 bound to cIAP2, only WT USP11 led to increased cIAP2 levels (Figures 2e and 3a). Inhibition of proteasomal degradation by MG132 notably stabilized cIAP2, whereas USP11 did not further stabilize cIAP2 upon MG132 treatment, suggesting that USP11 rescued cIAP2 from proteasomal degradation (Figure 3b). Moreover, the cIAP2 half-life was significantly delayed from 3 to 9 h in the presence of USP11 (Figure 3c), and ubiquitylation analysis suggested that WT USP11, but not MT USP11, decreased cIAP2 ubiquitylation (Figure 3d). In addition, USP11 almost completely removed the ubiquitin chain from ubiquitylated-cIAP2, resulting in free cIAP2 accumulation in vitro (Figure 3e). These results clearly supported the notion that USP11 is a direct cIAP2 DUB. Finally, ectopic USP11 expression abrogated exogenous cIAP2 degradation by BV6 (Figure 3f), suggesting that USP11 is a bona fide cIAP2 DUB that protects cIAP2 against proteasomal degradation.

Figure 3.

USP11 is a cIAP2-deubiquitylaing enzyme that stabilizes cIAP2. (a) 293FT cells were transfected with the plasmid expressing Myc-cIAP2 and increasing amounts of WT or catalytic inactive mutant (MT) FLAG-USP11. cIAP2 protein levels were determined by WB, and 2xGFP was used as a transfection control. (b) 293FT cells were transfected with cIAP2-expressing plasmid in the presence or absence of USP11 and treated with 10 μM MG132 for 6 h. (c) 293FT cells were transfected with the indicated plasmid and treated with 50 μg/ml CHX. Relative amounts of Myc-cIAP2 were calculated after normalizing to actin, and are shown in the right panel. (d) 293FT cells were transfected with the indicated plasmid and treated with MG132. Cells were lysed in 1% SDS and immunoprecipitated with an anti-Myc antibody, followed by WB using an HRP-conjugated anti-HA antibody. (e) Ubiquitylated cIAP2, purified by immunoprecipitation using HA-Ub, was incubated with FLAG-USP11 in the presence of ATP for 6 h. The cIAP2 ubiquitylation status was determined by WB using an anti-HA or anti-cIAP2 antibody. (f) 293FT cells were transfected with the cIAP2 expression vector with or without the FLAG-USP11 expression vector and treated with 2 μM BV6 for the indicated time period

USP11 inhibition decreases cIAP2 levels and sensitizes cells to apoptosis

We next tested whether USP11 is indeed involved in the regulation of cIAP2 stability by performing knockdown experiments. USP11 depletion using two independent siRNAs or lentiviral shRNAs resulted in cIAP2 downregulation in H1299 cells (Figure 4a; Supplementary Figure S3a). cIAP2 levels also decreased upon USP11 depletion in HeLa and MDA-MB-231 cells, suggesting that USP11 is a general cIAP2 regulator (Supplementary Figure 3b). Although cIAP1 and cIAP2 are highly homologous with redundant functions, USP11 did not regulate cIAP1 levels even though there was a slight interaction between the two in vitro (Figure 4a; Supplementary Figures S3a and c). Expression of other IAP members including XIAP, ML-IAP, and Survivin did not change upon USP11 depletion, suggesting that USP11 specifically controls cIAP2 levels (Figure 4a). In addition, cIAP2 mRNA levels did not change after USP11 depletion, suggesting that USP11 does not regulate cIAP2 at the transcriptional level (Figure 4b). MG132 treatment effectively rescued cIAP2 levels in USP11-depleted cells, which indicated that cIAP2 is degraded through the proteasome upon USP11 depletion (Figure 4c). Furthermore, endogenous ubiquitylated cIAP2 accumulated in USP11-depleted cells, suggesting that cIAP2 maintains its stability via USP11-dependent deubiquitylation (Figure 4d). Next, we assessed whether USP11 downregulation could restore the sensitivity of cIAP2 to BV6, leading to cell death. Indeed, USP11-depleted cells exhibited rapid cIAP2 degradation by BV6, which was comparable to cIAP1 degradation (Figure 4e). In accordance with decreased cIAP2 levels, complex II formation consisting of RIP1, FADD, and caspase-8 upon TNFα/BV6 treatment was increased, and caspase-8 was rapidly cleaved in USP11-depleted cells (Figure 4f; Supplementary Figure S3d). Finally, we found that USP11-depleted cells underwent rapid apoptosis following TRAIL/BV6 treatment (Figures 4g and h). Recently, mitoxantrone was identified as a USP11 inhibitor.³⁴ Consistently, USP11 inhibition by mitoxantrone induced cIAP2 destabilization, which was reversed with MG132 treatment (Figures 4h–i). USP11 depletion by siRNA abrogated mitoxantrone-induced cIAP2 degradation, suggesting that mitoxantrone regulates cIAP2 degradation via USP11 inhibition (Figure 4j). Furthermore, mitoxantrone sensitized cells to BV6- or birinapant-induced cell death (Supplementary Figure S3e). Taken together, our data indicate that USP11 is a positive cIAP2 regulator that prevents apoptotic cell death.

Figure 4.

USP11 knockdown results in cIAP2 downregulation and sensitizes H1299 cells to BV6-induced apoptosis. (a and b) H1299 cells were transfected with two independent siRNAs against USP11. Major IAP members were detected by WB (a), and cIAP2 and USP11 mRNA levels were determined by qRT-PCR (b). (c) H1299 cells transfected with USP11 siRNAs were treated with 10 μM MG132 for 12 h, and protein levels were determined by WB. (d) H1299 cells transfected as indicated were treated with MG132 for 12 h and lysed in 1% SDS. Cell lysates diluted in lysis buffer were immunoprecipitated with an anti-cIAP2 antibody. Ubiquitylated cIAP2 was detected by WB using an HRP-conjugated anti-Ub antibody on a PVDF membrane denatured with Guanidine-HCl. (e) USP11-knockdown H1299 cells were treated with the indicated concentration of BV6 for 30 min and analyzed by WB. (f) siRNA-transfected H1299 cells were treated with 20 ng/ml TNFα and 1 μM BV6. Cell lysates were immunoprecipitated with anti-caspase-8, followed by WB, to detect RIP1 and FADD. (g) H1299 cells transfected with the indicated siRNAs were treated with TRAIL and BV6 for 4 h. Cells were harvested, immediately stained with Annexin V and 7-AAD, and analyzed by flow cytometry. The Annexin V-positive/7-AAD negative fraction percentages are shown in dot plot data. (h) H1299 cells stably expressing USP11 shRNA were treated with TRAIL and BV6 for 6 h. Cell viability was determined by measuring the levels of ATP. (i–k) H1299 cells were treated with increasing concentration of mitoxantrone for 16 h (i) or 2 μM mitoxantrone in the presence of 10 μM MG132 for 16 h (j). H1299 cells were transfected with USP11 siRNA and treated with 2 μM mitoxantrone (k). cIAP1/2 proteins were detected by WB using the indicated antibodies

Previous reports have suggested that USP11 regulates TNFα-induced NF-κB activation through direct interaction with IKK or IκBα.^{35, 36} Because the NF-κB-dependent survival pathway is crucial for suppression of TNFα-dependent apoptosis, we monitored NF-κB activation and JNK activation in USP11-depleted cells. Consistent with previous reports, we found that USP11 depletion slightly enhanced IκBα phosphorylation and NF-κB target gene expression, such as TNFα and IL-8, in response to TNFα (Supplementary Figure S4a).^{35, 36} JNK activation and cIAP2 expression were similar in control and USP11-depleted cells (Supplementary Figure S4b). Despite the slight increase in NF-κB activity, which contributes cell survival, USP11 depletion eventually led to cell death via cIAP2 downregulation upon TNFα stimulation. Overall, depletion of USP11 in cells accelerated complex II formation and apoptotic cell death by destabilizing cIAP2.

USP11 is overexpressed in colon cancer and melanoma

We next investigated the function of USP11 in a more physiological context. Through mining of the Oncomine database, we found that USP11 is overexpressed in several cancers, including colorectal cancer and melanoma (Figures 5a and b). USP11 overexpression in colorectal cancer was also evident in another public data set (Supplementary Figure S5a). Furthermore, increased USP11 expression was detected in human melanoma metastases or colorectal cancers with relapse (Supplementary Figures S5b and c). Notably, we found that patients from various data sets with high USP11 levels exhibited poor prognosis (Figures 5c and d). On the basis of these data, we analyzed USP11 expression levels in several colon cancer and melanoma cell lines and found that the USP11 expression levels varied between cell lines (Figures 5e and f). For colon cancer cell lines, HCT116 cells showed high USP11 levels, whereas HCT-15 and HT-29 cells expressed USP11 at low levels (Figure 5e). We first tested these cells for cIAP2 status upon TNFα and BV6 treatment; when HCT-15 and HT-29 cells were pretreated with a low concentration of TNFα followed by BV6, cIAP1 and cIAP2 were rapidly degraded, whereas cIAP2 degradation was hindered in HCT116 cells (Figure 6a). These data suggest that USP11 overexpression may protect cancer cells against BV6 treatment. Supporting this hypothesis, USP11 depletion in HCT116 cells, but not in HCT-15 and HT-29 cells, sensitized cells to TNFα- and BV6-induced death processes (Supplementary Figure S6a). In the melanoma cell lines UACC-62 and UACC-257, cIAP2 was barely detectable but considerably increased following TNFα stimulation. Among these two cell lines, UACC-62 melanoma cells with high USP11 levels were protected against BV6-induced cIAP2 degradation and cell death compared with UACC-257 cells expressing low USP11 levels (Figure 6b; Supplementary Figure S6b). Because TNFα-induced cIAP2 expression potentially inhibits TRAIL-mediated apoptosis, we next investigated the roles of USP11 in TRAIL/BV6-mediated apoptosis in the presence of TNFα. USP11 knockdown significantly accelerated TRAIL/BV6-induced cell death in HCT116 cells, but not in HCT15 and HT-29 cells (Figure 6c; Supplementary Figure S6c). Consistently, USP11 depletion successfully restored BV6 sensitivity to TRAIL/BV6-induced apoptotic cell death in TNFα-pretreated UACC-62 cells (Figure 6d). Similar to BV6, birinapant showed limited cIAP2 degradation in HCT116 and UACC-62 cells, which express high USP11 levels (Supplementary Figures S7a and b). Upon USP11 depletion, cell death mediated by TNFα and birinapant was accelerated as expected (Supplementary Figures S7c and d).

Figure 5.

USP11 is overexpressed in colorectal cancer and melanoma and is associated with poor prognosis. (a and b) USP11 expression in colorectal cancer and melanoma was evaluating using data sets (GSE9348, GSE3189). Please refer to the Materials and methods for details. (c) Kaplan–Meier survival curves of colon cancer patients with high or low levels of USP11, according to the median expression level, from data sets (GSE28814). Statistical significance was calculated using Log rank (Mantel-Cox) test. (d) Kaplan–Meier survival curves of melanoma patients with high (top 30% of patients) or low (bottom of 30% patients) levels of USP11 from data sets (GSE22153). Statistical significance was calculated using Gehan-Breslow-Wilcoxon test. (e and f) USP11 expression in colon cancer and melanoma cell lines was analyzed by WB

Figure 6.

USP11 downregulation restores cell sensitivity to BV6 and enhances TRAIL and BV6 antitumor activity. (a and b) Colon cancer cell lines (a) and melanoma cell lines (b) were pretreated with 5 ng/ml TNFα for 12 h, followed by BV6 treatment for 30 min. cIAP protein levels were determined by WB. (c, d) HCT116 cells stably expressing shUSP11 (c) or UACC-62 cell lines transfected with siRNA (d) were treated with 2 ng/ml TNFα for 12 h, followed by TRAIL and BV6 stimulation for 8 h as indicated. Cell viability was determined by measuring the levels of ATP. (e) USP11-knockdown UACC-62 cells were pretreated with TNFα followed by TRAIL and BV6 treatment for 6 h. Caspase-8 and -3 cleavage was determined by WB. (f–h) HCT116 cells expressing shGFP or shUSP11 were injected into the flanks of nude mice. One week after inoculation, mice were treated with TRAIL (5 mg/kg) and BV6 (10 mg/kg) by i.p. injection as indicated. Tumor growth is shown in the graph (f), and tumors are shown in (g). Tumor mass is summarized in (h). Data represent the mean±S.D.; two-tailed Mann–Whitney test, ns=non-significant, **P<0.01 (n=7). (i–k) UACC-62 cells expressing shGFP or shUSP11 were injected into the flanks of nude mice. One week after inoculation, mice were treated with TRAIL (5 mg/kg) and BV6 (10 mg/kg) by i.p. injection. Tumor growth is shown in the graph (i), and tumors are shown in (j). Tumor mass is summarized in (k). Data represent the mean±S.D.; two-tailed Mann–Whitney test, ns=non-significant, **P<0.01, ***P<0.001 (n=6)

To verify the in vivo roles of USP11 in TRAIL/Smac mimetic-mediated tumor suppression, we used a xenograft tumor model consisting of HCT116 cells and UACC-62 cells expressing shGFP or shUSP11. These cells were injected into nude mice followed by BV6 and TRAIL treatment, and we found that the growth of HCT116/shUSP11 cell xenografted tumors was significantly slower upon treatment with TRAIL/BV6 compared with HCT116/shGFP cells (Figure 6f). The final mean tumor size and mass of control xenograft tumors treated with TRAIL/BV6 (631.5 mm³ and 0.62 g, respectively) were greatly decreased in USP11-depleted xenograft tumors (288.3 mm³ and 0.18 g, respectively) (Figures 6f–h). Similarly, UACC-62 melanoma xenografts also indicated that USP11 downregulation enhanced TRAIL/BV6 antitumor activity (700.6–289.3 mm³, 0.41–0.14 g) (Figures 6i–k). Taken together, these data imply that USP11 downregulation in colon cancer and melanoma effectively improves the tumor suppressive activity of TRAIL and BV6 in vivo.

TNFα stimulation stabilizes USP11 via the JNK pathway

We next evaluated whether USP11 contributes to cIAP2 accumulation upon TNFα treatment. The results showed that USP11 protein accumulated, with a 2.26-fold increase in mRNA levels, upon TNFα stimulation (Figures 7a and b). There was no obvious change in USP11 protein turnover rate in the presence or absence of TNFα, suggesting that USP11 expression is rather regulated at the transcriptional level (Supplementary Figure S8). While USP11-depleted cells exhibited similar cIAP2 mRNA levels, they failed to increase cIAP2 protein levels, suggesting that USP11 is responsible for cIAP2 maintenance after TNFα stimulation (Figures 7c and d). Next, we investigated the downstream pathway responsible for USP11 induction by treating cells with NF-κB, JNK, and p38 inhibitors. Among the three major pathways activated by TNFα signaling, we found that only JNK signaling suppressed USP11 transcription (Figures 7e and f). Consistent with previous reports, we found that JNK inhibition also reduced the protein and mRNA levels of cIAP2 (Figures 7e and f).³⁷ Interestingly, the JNK inhibitor SP600125 also decreased the half-life of cIAP2 protein, suggesting that JNK positively regulates cIAP2 expression by increasing transcription and USP11-mediated protein stability (Figure 7g). Consistent with the JNK-mediated changes in USP11 and cIAP2 expression levels, JNK inhibition accelerated TNFα or TNFα/BV6-induced apoptosis (Figure 7h). Thus, our data suggest that TNFα primarily prevents apoptosis by activating JNK-USP11-cIAP2, in addition to the NF-κB pathway; in contrast, JNK or USP11 inhibition downregulates cIAP2 expression and sensitizes cells to TNFα-induced apoptosis.

Figure 7.

USP11 is induced via the JNK pathway and stabilizes cIAP2 protein upon TNFα stimulation. (a) UACC-62 cells were treated with 10 ng/ml TNFα for the indicated time period, and USP11 and cIAP2 expression was determined by WB. (b) cIAP2 and USP11 transcript levels in TNFα-treated UACC-62 cells were measured by qRT-PCR. (c and d) UACC62 cells expressing shUSP11 were treated with TNFα as indicated. USP11 and cIAP2 protein and mRNA levels were analyzed by WB (c) and qRT-PCR (d), respectively. (e) UACC-62 cells were pretreated with a JNK inhibitor (SP600125, 20 μM), p38 MAPK inhibitor (SB203580, 20 μM), or the NF-kB inhibitor IV (10 μM) 30 min before TNFα treatment. Twenty-four hours after TNFα treatment, cell lysates were analyzed by WB. (f) UACC-62 cells were treated with TNFα in the presence of SP600125. mRNA levels were determined by qRT-PCR. (g) UACC-62 cells were pretreated with TNFα in the absence or presence of SP600125 for 12 h. Cells were treated with 50 μg/ml CHX for the indicated time period, and cIAP2 stability was determined by WB. Relative amounts of cIAP2 were calculated after normalizing to actin, and are shown in the right panel. (h) UACC-62 cells were treated with a combination of TNFα, BV6, and SP600125 for 24 h. Cell viability was determined by measuring the levels of ATP. (i) Schematic model of USP11-mediated cIAP2 regulation following TNFα stimulation. (Data represent the mean±S.D.; n=3, *P<0.05, **P<0.01, ***P<0.001)

Discussion

Cancer cells express several IAPs, including cIAP1/2, XIAP, and Survivin, to evade apoptosis induced by various stimuli.^{4, 5, 38} Among these, cIAP1 and 2 are highly homologous and have been considered functionally redundant.⁴ Most biological studies have focused on cIAP1 rather than cIAP2 because cIAP1 is more highly expressed under normal conditions, suggesting its predominant role.³⁹ In contrast, the absence of both cIAP1 and cIAP2 is required for complete inhibition of TNFα-mediated survival, suggesting that these inhibitors have redundant functions.^{20, 21} Evidence also supports that cIAP2 has independent roles from cIAP1 in suppressing apoptosis. For examples, we and others have found that the levels of cIAP2, but not cIAP1, are greatly increased after TNFα stimulation. cIAP2 is also induced by IL-1β and cisplatin, leading to apoptosis resistance.^{40, 41} In addition to these observations, elevated cIAP2 expression in certain cell lines renders cells resistant to Smac mimetics, suggesting that cIAP2 is the decisive factor for protecting cancer cells from Smac mimetic-induced apoptosis.²²

In this study, we show that a USP11-dependent mechanism selectively and specifically regulates cIAP2 stability independent of cIAP1, which may explain the diverse responses of cancer cells to treatment with Smac mimetics, which are currently being tested in the clinic (Figure 7i).^5,14 In contrast to cIAP1, the cIAP2 degradation rate upon Smac mimetic treatment largely varied between cell lines and correlated with USP11 expression. For example, cIAP2 expression in cells with high USP11 levels such as H1299, HCT116, and UACC-62 was associated with strong resistance against Smac mimetic treatment. When USP11 was depleted in these cell lines, BV6 or birinapant induced cIAP2 degradation markedly increased, with a notable increase in TNFα or TRAIL-induced apoptosis. The results reported here suggest that USP11-induced cIAP2 stabilization may be a decisive factor for determining cellular fate upon treatment with Smac mimetics.

Given their critical roles in apoptosis inhibition, IAPs are attractive therapeutic targets. While the Smac mimetics can induce apoptosis through autocrine TNFα, cancer cells often overcome Smac mimetic-induced antitumor activities. Therefore, other strategies to stimulate cell death pathways have been studied to overcome Smac mimetic resistance. Because IAPs primarily control the extrinsic apoptosis pathway, combination treatment with TRAIL synergistically enhanced their antitumor activity in xenograft experiments in various cancer models.^{16, 42} In addition, concomitant inhibition of receptor tyrosine kinases (RTKs), which are often activated by Smac mimetics, successfully enhanced the antitumor effects in a xenograft model.^{29, 43} Recently, combination treatment with a demethylating agent largely improved the tumor-suppressive effects of Smac mimetics in several cancer models.^{27, 44} In addition to these, numerous new compounds including LCL161 and birinapant are currently in phase I and II clinical trials.^{5, 14} As our data suggest, cIAP2 sensitivity against Smac mimetics in relation to USP11 expression should be considered in the abovementioned combination therapies. Using specific shRNAs, we showed that USP11 downregulation in colon cancer and melanoma efficiently suppressed tumor growth in xenograft models. In addition, the type II topoisomerase inhibitor mitoxantrone, which is widely used for the treatment of various tumor types, was recently identified as an USP11 inhibitor. Although mitoxantrone involves various pathways, such as inhibition of topoisomerase and induction of immunogenic cell death, to suppress tumor growth,^{45, 46, 47} our data suggest that inhibition of the USP11-cIAP2 pathway by mitoxantrone contributes to its anticancer effects. Thus, combination treatment with mitoxantrone and Smac mimetics and/or TRAIL may synergistically enhance the desired antitumor effects.

Finally, we found that USP11 expression was increased in response to TNFα, implying the relevance of USP11 function in TNFα signaling (Figure 7i). TNFα triggers a variety of downstream signaling pathways such as NF-κB, JNK, and p38. In particular, NF-κB activation following TNFα stimulation induces the expression of several pro-inflammatory and pro-survival genes. TNFα-mediated JNK signaling activates a wide array of genes via the AP-1 transcription factor. Although cIAP2 has been identified as an NF-κB target, it is also upregulated by the JNK-AP-1 pathway.³⁷ Supporting these data, we found that cIAP2 induction upon TNFα stimulation was suppressed following treatment with a JNK inhibitor. Interestingly, this JNK inhibitor caused a decrease in USP11 mRNA and protein levels, suggesting that the JNK pathway is responsible for cIAP2 and USP11 induction. In addition, TNFα/JNK-dependent USP11 induction may serve as a major cIAP2-stabilizing factor, as cIAP2 accumulation following TNFα stimulation was suppressed in USP11-depleted cells. Because JNK is involved in various signaling pathways, its roles in apoptosis remain controversial and likely depend on different cellular contexts. JNK activation primarily leads to apoptosis by increasing pro-apoptotic gene transcription or by directly activating mitochondrial apoptosis.^{48, 49} However, JNK inhibition by SP600125 can also sensitize cells to death in certain conditions, suggesting that JNK may have anti-apoptotic functions.^{50, 51} Consistently, our data have shown that JNK inhibition downregulated both USP11 and cIAP2, thereby accelerating TNFα- or TNFα/BV6-induced cell death. Our results further suggest that JNK may represent a clinical target along with TRAIL and Smac mimetics. To summarize, our studies identified a novel survival axis involving TNFα/JNK/USP11/cIAP2, which may function to protect cells from extrinsic apoptotic stimuli. On the basis of these observations, USP11 and cIAP2 should be considered therapeutic markers and targets. Targeting these molecules may serve as an effective tumor treatment in combination with TRAIL and Smac mimetics.

Materials and methods

Plasmids

Sixty-seven FLAG-HA-DUB constructs were purchased from Addgene (http://www.addgene.org).⁵² The pCMV-FLAG-USP11 vector was kindly provided by JH Chung (Seoul National University, Seoul, Korea) and was subcloned into pGEX-4T-1. Myc-cIAP2 constructs were kindly provided by SY Lee (Ewha Woman's University Ewha, Seoul, Korea), and pcDNA3-His-Ub was kindly provided by DP Lane.⁵³ pHM6-HA-Ub has been described previously.⁵⁴ The pEGFP-C2 (Clontech, San Diego, CA, USA) and pCS3-MT-6XMyc vectors were used as transfection controls.

Antibodies and chemicals

The following antibodies were used: cIAP2 (3130; Cell Signaling, Danvers, MA, USA, AF8171; R&D Systems, Minneapolis, MN, USA), cIAP1 (AF8181; R&D Systems), XIAP (2045; Cell Signaling), ML-IAP (5471; Cell Signaling), Survivin (2808; Cell Signaling), USP11 (sc-365528; Santa Cruz, Santa Cruz, CA, USA), Caspase-8 (9746; Cell Signaling, sc-6136; Santa Cruz), Caspase-3 (9662; Cell Signaling), hFADD (610399; BD Transduction Laboratories, Franklin lakes, NJ, USA), RIP1 (610409; BD Transduction Laboratories), IκBα (4814; Cell Signaling), Phospho-IκBα (Ser32) (2859; Cell Signaling), NF-κB p65 (sc-8008; Santa Cruz), Phospho-NF-κB p65 (Ser536) (3033; Cell Signaling), SAPK/JNK (9258; Cell Signaling), Phospho-SAPK/JNK (Thr183/Tyr185) (9251; Cell Signaling), HRP-conjugated-α-Ub FK2 (PW0150; Biomol-Enzo, Farmingdale, NY, USA), HA (sc-805 rabbit; Santa Cruz, 3F10 rat; Roche, Mannheim, Germany, 12013819001), FLAG (F3165 mouse, F7425 rabbit; Sigma-Aldrich, St. Louis, MO, USA), GST (sc-138; Santa Cruz), GFP (sc-8334; Santa Cruz), Myc (sc-40; Santa Cruz), β-actin (A5316; Sigma), normal mouse IgG (sc-2025; Santa Cruz), normal rabbit IgG (sc-2027; Santa Cruz), and normal goat IgG (sc2028; Santa Cruz). BV6 was kindly provided by D Vucic (Genentech, San Francisco, CA, USA). pQE-His-TRAIL was a kind gift from SH Kim (Korea National Cancer Center, Seoul, Korea) and was used for the purification of His-tagged human TRAIL and for in vivo experiments as previously reported.⁵⁵ CHX (C4859), N-ethylmaleimide (NEM, E3876), human recombinant TNFα (T0157), ActD (A4262), and dimethyl sulfoxide (DMSO) were purchased from Sigma-Aldrich. MG132 was purchased from A.G. Scientific (San Diego, CA, USA). Mitoxantrone (S1889), SP600125 (JNK inhibitor, S1460), and SB203580 (p38 MAPK inhibitor, S1076) were purchased from Selleck Chemicals (Huston, TX, USA). NF-κB Activation Inhibitor IV (481412) was purchased from Merck (Darmstadt, Germany).

Cell culture, transfection, and cell viability analysis

H1299 (human lung carcinoma cell line), HeLa (cervical cancer cell line), 293FT (human embryonic kidney), and MDA-MB-231 (breast cancer cell line) cells were maintained in Dulbecco's modified Eagle's medium (DMEM, Corning Cellgro, Manassas, VA, USA) supplemented with 10% fetal bovine serum (Corning Cellgro) and 1% penicillin/streptomycin (Invitrogen, Carlsbad, CA, USA) in 5% CO₂ at 37 °C. Colo205, HCC-2998, HCT116, HCT15, HT-29, KM12, SW620, SW480, and RKO cells (human colon cancer cell lines) and LOX-IMVI, MALME-3M, M14, MDA-MB-435, SK-MEL-2, SK-MEL-5, SK-MEL-28, UACC-62, and UACC-257 cells (melanoma cell lines) were maintained in RPMI with 10% FBS and antibiotics (GIBCO, Grand Island, NY, USA). PEI (Sigma-Aldrich) was used to transfect 293FT cells, and Lipofectamine 2000 was used to transfect 293FT cells for lentivirus production. To determine cell viability, cells were incubated with Cell Titer Glo reagent and analyzed with a luminometer according to the manufacturer's protocol (Cell Titer Glo Luminescent Cell Viability Assay kit, G7571, Promega, Madison, WI, USA).

Immunoprecipitation and DISC immunoprecipitation

Cells were lysed in a buffer composed of 50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 0.5% Triton X-100, and 1 mM EDTA containing a protease inhibitor cocktail. The cell lysates were immunoprecipitated by incubation with antibodies for 2 h, followed by incubation with protein G sepharose (GE Healthcare, Buckinghamshire, UK) for 2 h. Then, the immunoprecipitates were boiled in 2X sample buffer for 5 min. For DISC immunoprecipitation, cells were treated with TNFα (20 ng/ml) and BV6 (1 μM) in the presence of zVAD-fmk (20 μM) for 1 or 2 h. Cells were lysed in DISC immunoprecipitation buffer containing 50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1% Triton X-100, 10% glycerol, and 1 mM EDTA with a proteasome inhibitor cocktail. DISC complexes were purified by immunoprecipitation using an α-Caspase-8 goat (sc-6136) antibody, followed by WB analysis using α-FADD mouse (610399), α-Caspase-8 mouse (#9746), and α-RIP1 mouse (610409) antibodies.

Protein purification and in vitro binding assays

Recombinant human USP11 protein was expressed in Escherichia coli using the pGEX-4T-1 plasmid and purified using glutathione sepharose 4B (GE Healthcare). Recombinant human cIAP1 (818-IA) and cIAP2 (817-P2) were purchased from R&D Systems. Recombinant cIAP1 or cIAP2 was incubated with GST or GST-USP11 for 2 h, and then glutathione sepharose was added to the mixture and incubated for 1 h. GST complexes were washed and eluted in 10 mM reduced glutathione. Proteins were analyzed by WB.

In vivo and in vitro deubiquitylation assays

The ubiquitylation assay was performed as described previously.⁵⁶ Briefly, cells were lysed in 6 M guanidine-HCl buffer (pH 8) containing 5 mM NEM (Sigma-Aldrich), and His-ubiquitin-conjugated proteins were purified by Ni²⁺-NTA (Qiagen, Valencia, CA, USA) and analyzed by western blotting (WB). For ubiquitylation assays using HA-Ub, cells were lysed under denaturing conditions containing 1% SDS. Cell lysates were immunoprecipitated, followed by WB with a horseradish peroxidase (HRP)-conjugated anti-HA antibody (12013819001, Roche). For the detection of endogenous cIAP2 ubiquitylation, cells were treated with 10 μM MG132 for 12 h and harvested in PBS containing NEM and lysed in 1% SDS. Cell lysates were immunoprecipitated with anti-cIAP2 antibodies (3130, Cell Signaling) in lysis buffer containing 0.1% SDS, followed by WB. Transferred proteins were denatured using 6 M guanidine-HCl, followed by WB using an HRP-conjugated anti-Ub antibody (FK2, PW0150, Biomol-Enzo). For in vitro ubiquitylation, ubiquitylated Myc-cIAP2 was obtained from 293FT cells transiently expressing Myc-cIAP2 and HA-Ub by immunoprecipitating cIAP2 using anti-Myc antibody as described above. FLAG-USP11 was purified from 293FT cells expressing FLAG-USP11 using ANTI-FLAG M2 Affinity Gel (A2220, Sigma) and eluted with FLAG peptide (F3290, Sigma) according to the manufacturer's protocol. Ubiquitylated cIAP2 was incubated with purified FLAG-USP11 in deubiquitylation buffer containing 40 mM Tris-HCl (pH 7.6), 50 mM NaCl, 5 mM MgCl2, 5 mM DTT, and 2 mM ATP for 6 h. The ubiquitylation status of cIAP2 was detected by WB using HRP-conjugated anti-HA antibody or anti-cIAP2 antibody.

USP11 knockdown using siRNA and shRNA

Two USP11 flexitube siRNAs (siUSP11#4, (AC)CGATTCTATTGGCCTAGTA and siUSP11#5, (CT)GCGTCGGGTACGTGATGAA) were purchased from Qiagen and resynthesized. Control siRNA (siCon, 1027281) was also purchased from Qiagen. The ON-TARGET plus SMARTpool for human cIAP1 siRNA (L-004390-00), human cIAP2 siRNA (L-004099-00), and a non-targeting pool (siNON, D-001810-10) were purchased from Dharmacon/Thermo Scientific (Waltham, MA, USA). Cells were transfected with siRNAs using Lipofectamine and RNAiMax (Invitrogen) via the reverse-transfection method according to the manufacturer's protocol. Mission lentiviral shRNA expression vectors for human USP11 (shUSP11#2, TRCN0000007359, CCGGCCGTGATGATATCTTCGTCTACTCGAGTAGACGAAGATATCATCACGGTTTTTG, and shUSP11#4, TRCN0000315154, CCGGCCGTGACTACAACAACTCCTACTCGAGTAGGAGTTGTTGTAGTCACGGTTT TTG) and control GFP were purchased from Sigma-Aldrich. 293FT cells were transfected with pLKO.1 shRNA-expressing vector and packaging vectors for 48 h, and supernatants were harvested, filtrated, and transferred to H1299, HCT116, and UACC-62 cells. shRNA-expressing cells were selected following 1–2 μg/ml puromycin treatment for 7 days.

qRT-PCR analysis

Total RNA was extracted with Trizol reagent (Invitrogen) according to the manufacturer's instructions. cDNA was synthesized from 1 μg total RNA using PrimeScript Reverse Transcriptase (2680A; Takara Bio, Shiga, Japan), amplified, and analyzed with the QuantiTect SYBR Green PCR Kit followed by qRT-PCR (Qiagen) with the following primers:

USP11, 5′-CAAGAACAAGGTTGGCCATT-3′ (forward) and 5′-TCGCACAGCTCCACATACTC-3′ (reverse)

cIAP2, 5′-ATTTGATGAAAAGCGCCAAC-3′ (forward) and 5′-AAACCAGCACGAGCAAGACT-3′ (reverse)

TNFα, 5′-AACCTCCTCTCTGCCATCAA-3′ (forward) and 5′-GGAAGACCCCTCCCAGATAG-3′ (reverse)

IL-8, 5′-GTGCAGTTTTGCCAAGGAGT-3′ (forward) and 5′-AAATTTGGGGTGGAA AGG TT-3′ (reverse)

GAPDH, 5′-GATCATCAGCAATGCCTCCT-3′ (forward) and 5′-TGTGGTCATGAGTCCTTCCA-3′(reverse)

Analysis of USP11 expression in various cancers using public data sets

USP11 expression was analyzed in various tumor types using the Oncomine database (www.oncomine.org)⁵⁷ (GSE9348,⁵⁸ GSE3189⁵⁹). USP11 expressions at various stages of colorectal cancer and melanoma were further confirmed using Gene Expression Omnibus (GEO) profiles (www.ncbi.nlm.nih.gov/geoprofiles) (GSE4107,⁶⁰ GSE18088,⁶¹ and GSE8401⁶²). The PROGgene database (www.compbio.iupui.edu/proggene) was used to identify the data sets for survival analysis and re-analyzed the public GEO data sets (www.ncbi.nlm.nih.gov/gds) (GSE28814).⁶³ The melanoma survival rate was directly analyzed using the other GEO data set (GSE22153).⁶⁴

Xenograft study

Approximately 2 × 10⁶ HCT116 or 1 × 10⁶ UACC-62 stable cells expressing shGFP or shUSP11 were resuspended in 50 μl PBS and 50 μl Matrigel matrix (354234, BD Biosciences, Franklin Lakes, NJ, USA) and subcutaneously inoculated into 6-week-old female Balb/c nude mice (Narabiotech, Korea). One week after inoculation, mice bearing xenograft tumors were treated with TRAIL (5 mg/kg, every 2 days, total eight treatments for HCT116 cells and total six treatment for UACC-62 cells) and BV6 (10 mg/kg, every 4 days, total four treatments for HCT116 cells and total three treatment for UACC-62 cells) by intraperitoneal (i.p.) injection. Tumor volume was measured using calipers and calculated according to the following formula: volume=length × width² × 0.5236. All animal experiments were performed in specific pathogen-free facilities and in accordance with the Guidelines for The Institutional Animal Care and Use Committees of the Laboratory Animal Research Center at Yonsei University (Korea). To assess whether differences between groups were significant, a two-tailed Mann–Whitney test was performed using GraphPad Prism version 5.01 (GraphPad Software, La Jolla, CA, USA).

Statistics

Statistical significance was assessed by Student's t test or a two-tailed Mann-Whitney test using GraphPad Prism version 5.01.

Glossary

IAP

inhibitor of apoptosis

cIAP2

cellular IAP2

BIR

baculoviral IAP repeat

XIAP

X chromosome-linked IAP

ML-IAP

melanoma IAP

USP11

ubiquitin-specific protease 11

TNFR

tumor necrosis factor receptor

TRAF2

TNFα receptor-associated factor 2

RIPK1

receptor interacting protein kinase

TRAIL

TNFα-related apoptosis-inducing ligand

TLR

Toll-like receptor

Smac

second mitochondria-derived activator of caspases

PI3K

phosphoinositide 3-kinase

TGFβ

transforming growth factor-beta

The authors declare no conflict of interest.