IGF2BP1 controls cell death and drug resistance in rhabdomyosarcomas by regulating translation of cIAP1

INTRODUCTION

Rhabdomyosarcoma (RMS) is the most common soft tissue tumour in children and represents 3–4% of all childhood cancers.1 RMS are malignant tumours of the muscle typified by myoblast-like cells that have lost the capacity to fully differentiate.2 If treated aggressively, RMS patients have a high long-term survival rate; however, chemotherapeutic resistance remains a large problem.1–3 Understanding the molecular basis of this resistance, therefore, provides an opportunity for targeted drug therapies.

The failure of current chemotherapeutic approaches to eradicate cancer cells is frequently owing to defects in the execution of cellular apoptotic programme.4 Apoptosis is the mechanism by which multicellular organisms orchestrate death and removal of damaged cells, thus maintaining tissue home- ostasis. It is tightly regulated by pro- and antiapoptotic factors. Among these factors, cellular inhibitor of apoptosis 1 (cIAP1, also known as HIAP2, MIHB), a member of the IAP family, is a key regulator of apoptosis and promotes cancer cell survival by controlling the nuclear factor-κB signalling and extrinsic cell death pathways (reviewed in Gyrd-Hansen and Meier5).

Expression of cIAP1 is tightly regulated at the level of protein synthesis through an internal ribosome entry site (IRES).6–8 IRES elements are discrete RNA elements found in the 5′-untranslated region (UTR) of a number of viral and cellular mRNAs that facilitate recruitment of the ribosome to the translation initiation start independently of the 5′ cap.9 The mechanisms by which cellular IRES elements mediate ribosome recruitment are still not fully understood. However, it has been shown that in addition to RNA structure and some canonical translation initiation factors, other cellular proteins are required for proper IRES function.10,11 These proteins, termed IRES trans-acting factors (ITAFs), are thought to help in the recruitment of the ribosome by acting either as scaffold proteins or RNA chaperones.12 We have previously identified four potential ITAFs interacting with cIAP1 IRES, among which was the insulin-like growth factor 2 mRNA-binding protein 1 (IGF2BP1).8

IGF2BP1 (also known as IMP1 or ZBP1) is a member of the VICKZ family of RNA-binding proteins13 that was first identified in the human RMS cell line RD, along with its paralogues IGF2BP2 and IGF2BP3, as factors that bind to the human IGF-II leader-3 mRNA and regulate its translation.14 IGF2BP1 is emerging as a key regulator of mRNA metabolism with diverse role in the control of mRNA localisation,15 stabilisation16,17 and translation.14,18–20 Importantly, IGF2BP1 is an oncofetal protein that is normally only expressed during embryogenesis21 but is re-expressed in a variety of cancers (reviewed in Yisraeli13 and Bell et al.22). Although only one study has directly linked IGF2BP1 expression to tumour- igenesis in vivo,23 there is substantial evidence in vitro pointing to its oncogenic potential.14,17–19,24–26

Here, we report that IGF2BP1 is overexpressed in primary human RMS tumours and cell lines where it drives high expression of cIAP1 by enhancing IRES-mediated translation of cIAP1. Importantly, reducing the levels of cIAP1 in RMS cell lines, either by IGF2BP1 knockdown or by Smac mimetic compound (SMC) treatment, sensitises RMS cells to tumour necrosis factor-α (TNFα)-mediated cell death. Finally, targeting cIAP1 by SMC inhibits the establishment and growth of RMS xenograft tumours in mice. Our results identify IGF2BP1 as a critical regulator of cIAP1 expression and apoptotic resistance in RMS and advocate for the combined use of IAP antagonists and TNFα as a potential therapeutic approach for RMS.

RESULTS

IGF2BP1 is overexpressed in RMS and drives cIAP1 expression We initially identified IGF2BP1 as one of the proteins that specifically binds to the cIAP1 IRES.8 As IGF2BP1 was first discovered in an RMS cell line and has been reported to be overexpressed in a variety of tumours,13,22 we decided to investigate a possible link between IGF2BP1 and cIAP1 expression in RMS tumours and tumour-derived cell lines. We observed a striking pattern of IGF2BP1 overexpression (10–60-fold) in established RMS cell lines of different origin (alveolar and embryonic) compared with normal human skeletal muscle myoblasts (HSMM) (Figure 1a and Supplementary Figures 1a and b). Notably, cIAP1 showed the same pattern of expression, being overexpressed 10–32-fold, compared with HSMMs (Figure 1a and Supplementary Figures 1a and b). Elevated IGF2BP1 and cIAP1 expression was also noted in three out of four primary human embryonic RMS tumours when compared with human fetal and adult skeletal muscle (Figure 1b). In addition, formalin-fixed, paraffin-embedded primary human RMS tumours were analysed by immunohistochemistry and compared with normal human skeletal muscle sections (Figure 1c). IGF2BP1 was moderately to highly overexpressed (score 2 or 3) in six out of eight RMS cases analysed compared with normal muscle sections (score 1; seven out of nine cases), and this correlated well with increased cIAP1 expression (Figure 1d, Spearman’s correlation P-value of 0.033; Supplementary Figures 1c and d).

We also examined the expression of two other members of the IAP family, cIAP2 and X-linked IAP (XIAP), in RMS cell lines, and found that while cIAP2 protein was barely detectable, XIAP protein was not significantly increased, except for RH30 cells (Supplementary Figures 1a and b).To demonstrate that high expression of IGF2BP1 drives cIAP1 expression in RMS, we knocked down IGF2BP1 in the RH36 cell line and observed a concomitant decrease in cIAP1 protein levels compared with cells transfected with a non-targeting control small interfering RNA (siRNA) (Figure 1e). Interestingly, the reduction in cIAP1 protein levels was not associated with changes in cIAP1 steady-state mRNA levels (Figure 1f). Furthermore, while IGF2BP1 steady-state mRNA levels were elevated in all six RMS cell lines, cIAP1 mRNA levels were significantly decreased (Supplementary Figure 1e), arguing for a posttranscriptional control of cIAP1 expression. Collectively, these data and the fact that we have previously identified IGF2BP1 as a cIAP1 IRES-binding protein prompted us to hypothesise that IGF2BP1 is an IRES trans-acting factor that regulates cIAP1 IRES-mediated translation.

IGF2BP1 regulates cIAP1 translation

We next set out to characterise the mechanism by which IGF2BP1 regulates cIAP1 expression. For these experiments, we used human embryonic kidney 293 (HEK293) cells that recapitulate the IGF2BP1-mediated changes in cIAP1 expression seen in RMS cells (Figures 2a and b). Importantly, re-expression of IGF2BP1 in cells previously transfected with IGF2BP1 siRNA was able to rescue cIAP1 protein to levels comparable with that of control siRNA- treated cells (Figure 2c). In addition, as IGF2BP1 knockdown did not affect cIAP2 and XIAP expression at the mRNA or protein levels (Supplementary Figures 1f and g), these results demonstrate that IGF2BP1 specifically regulates cIAP1 protein levels.

To demonstrate that IGF2BP1 regulates cIAP1 translation, we performed polysome profiling to examine the association of endogenous cIAP1 mRNA with translating ribosomes upon IGF2BP1 knockdown (Figure 2d). Real-time reverse transcription- –polymerase chain reaction (RT–qPCR) amplification of cIAP1 mRNA from individual ribosome fractions showed a decreased association with polysomes in cells with reduced IGF2BP1 levels, as shown by a decrease in cIAP1 mRNA polysome to monosome ratio (Figure 2e and Supplementary Figure 2a). In contrast, association of RPL13A, a representative housekeeping mRNA, with polysomes remained unchanged (Figure 2e and Supplementary Figure 2b). Importantly, IGF2BP1 knockdown did not affect the overall polysomes profiles except for a slight decrease in total polysome quantities (Figure 2d and Supplementary Figure 2c), which could be explained by reported association of IGF2BP1 with more than 1000 transcripts and possibly within polysomes.27,28 These results are consistent with the notion that IGF2BP1 directly regulates translation of the cIAP1 mRNA.

IGF2BP1 directly binds cIAP1 mRNA and mediates its translation via the 5′-UTR IRES

As our data showed that IGF2BP1 controls translation of cIAP1 mRNA, we questioned whether this control is exerted through the 5′-UTR of cIAP1, and specifically through its IRES. We first tested this possibility using a monocistronic reporter in which the cIAP1 5′-UTR controls translation of the downstream CAT reporter gene.

We observed that knocking down IGF2BP1 resulted in a significant decrease in CAT expression from the cIAP1-5′-UTR-controlled reporter (cIAP1-5′-UTR-pMC) when compared with non-targeting siRNA-transfected cells, whereas CAT expression from the empty reporter plasmid (pMC) remained unaffected (Figure 3a). Importantly, this decrease in CAT expression was not due to changes in CAT mRNA levels (Figure 3a, bottom), indicating that IGF2BP1 regulates cIAP1 translation through the 5′-UTR. To further determine if IGF2BP1 regulates cIAP1 by the IRES-mediated translation, we performed a bicistronic reporter assay in which β-galactosidase (β-gal) expression reports cap-dependent translation, whereas CAT expression is driven by the cIAP1 IRES (pBC-cIAP1-5′-UTR; Graber et al.8). We found that cIAP1 IRES activity was specifically decreased in cells expressing IGF2BP1 short hairpin RNA (shRNA) but not in cells expressing the scrambled shRNA, or when a bicistronic reporter containing the inverted cIAP1-5′-UTR was used (Figure 3b). Importantly, neither IGF2BP1 knockdown nor the directionality of the cIAP1 IRES affected β-gal expression (Figure 3b, bottom), indicating that regulation of CAT expression through the IRES is separate from that of β-gal, directed by the cap.

We next wished to examine whether IGF2BP1 binds directly to the cIAP1 IRES. Ultraviolet crosslinking of incubated GST-IGF2BP1 and 32P-labelled cIAP1 IRES RNA probes showed that GST-IGF2BP1 directly binds to the cIAP1 full-length IRES (probe 1) and to probes corresponding to the 3′ end (probes 2 and 5) but not the 5′ end of the IRES (probes 3 and 4) (Figure 3c). To further delineate IGF2BP1- binding region(s), we searched the cIAP1 IRES sequence for IGF2BP1 putative binding sites (GGACU/ACACC29 or CAUH, where H = A, U or C28) and identified two sites that fit these criteria (sites A and B; Figure 3d). Although deletion of site A (probe 5-ΔA) did not abrogate binding of GST-IGF2BP1, deletion of site B (probe 5-ΔB) or both sites (probe 5-ΔAB) resulted in a loss of binding (Figure 3e, left panel). However, when tested in the context of the full-length IRES (probe 1), site B only accounted for part of GST-IGF2BP1 binding (Figure 3e, right panel). Further mapping revealed that GST-IGF2BP1 binds to probe 6 (–150 to − 43 of cIAP1 IRES) but not to probe 3 (–150 to − 64) (Figure 3f), suggesting that GST-IGF2BP1 binds to the region containing stemloops (SLs) II and III (Figure 3d). To test this possibility, we generated sequential deletions of this region and found that deletion of site C-containing SLIII (Figure 3d, probe 1-ΔC) resulted in a significant loss of binding, whereas double deletion of sites B and C (probe 1-ΔBC) resulted in a complete loss of binding (Figure 3g).

These results show that GST-IGF2BP1 binds to SLIII of the cIAP1 IRES and to the ‘ACAUUA’ site (site B) proximal to the AUG start codon. To correlate GST-IGF2BP1 binding with cIAP1 IRES activity, we constructed pBGal/CAT bicistronic plasmids in which sites B and C of the cIAP1 minimal IRES are deleted individually or in combination. As expected, deletion of site B did not result in changes of cIAP1 IRES activity; however, the contribution of site C to cIAP1 IRES activity could not be confirmed as the construct harbouring this deletion was not functional (data not shown). Altogether, these results suggest that IGF2BP1 binding to the stemloop III is critical in mediating cIAP1 IRES activity and function in cells.

IGF2BP1 knockdown sensitises RMS cells to TNFα-mediated cell death
IGF2BP1 has been emerging as an important oncogenic factor, in particular with respect to cellular proliferation and metastasis, through the regulation of proto-oncogenes such as MYC and KRAS.17,24 In contrast, the role of IGF2BP1 in controlling apoptosis in cancer cells has not been investigated. As cIAP1 is a key regulator of caspase-8-mediated cell death5 and, as we have now shown, a posttranscriptional target of IGF2BP1, we asked whether elevated levels of IGF2BP1 have a role in RMS cell survival. Consequently, we examined the effect of IGF2BP1 knockdown on RH36 cells treated with TNFα, as recent reports suggested that this RMS-derived cell line (among others) is resistant to TNFα-related apoptosis-inducing-ligand (TRAIL).30–32
Treatment of RH36 cells with increasing concentrations of TNFα did not affect their viability (Figure 4a). However, IGF2BP1 knockdown (Supplementary Figure 2e) significantly reduced RH36 cell viability when compared with non-targeting siRNA- transfected cells (Figure 4a). Similarly, IGF2BP1 knockdown significantly increased caspase-3/7 activity upon TNFα treatment, compared with untreated cells (Figures 4c and d). Importantly, IGF2BP1 knockdown alone did not affect RH36 cell viability or caspase activity (Figures 4b–d), indicating that reducing IGF2BP1 levels does not cause general toxicity but instead sensitises RH36 cells to TNFα-mediated death. Importantly, restoring cIAP1 expression by Adv-HIAP2 reduced caspase-3/7 activity in IGF2BP1 siRNA-transfected cells to levels comparable with that of non-targeting siRNA-transfected ones, whereas a control LacZ adenovirus did not have the same effects (Figures 4e and f). These results show that IGF2BP1 controls TNFα-mediated cell death through the regulation of cIAP1 translation.

cIAP1 depletion by SMCs sensitises RMS cells to TNFα-mediated cell death

We have shown that cIAP1 is the key factor downstream of IGF2BP1, which mediates RMS sensitivity to TNFα. To test the therapeutic utility of our observations, we set out to target directly cIAP1 by the use of SMCs. SMC treatment triggers the autoubiquitination and proteasomal degradation of cIAPs, leading to the activation of the non-canonical NF-κB pathway, activation of caspase-8 and cell death. SMCs can also antagonise XIAP inhibition of caspases through the binding to the BIR2 domain.33–35 We observed that cIAP1 depletion from RH36 cells by the dimeric SMC AEG40730 (Bertrand et al.36 Supplementary Figure 2f) greatly reduced RH36 viability in the presence of TNFα (Figures 5a, c and d) and increased caspase-3/7 activity by threefold compared with vehicle-treated cells (Figures 5e and f). Importantly, SMC treatment alone did not affect the viability of RH36 cells (Figures 5b–f). We also found that AEG40730 sensitises RH36 cells to TRAIL-mediated cell death (Supplementary Figure 3), as was reported previously.3,37,38 Importantly, AEG40730 sensiti- sation to TNFα- and TRAIL-mediated cell death was also observed in another RMS cell line, RH41 (Supplementary Figure 4).

Because SMC can also target cIAP2 for degradation and inhibit XIAP activity, we examined their contribution to the sensitisation of RH36 cells to TNFα. cIAP2 is not expressed in RH36 cells (Supplementary Figure 1a) and XIAP knockdown did not significantly sensitise RH36 to TNFα-mediated cytotoxicity com- pared with untreated cells, in stark contrast to cIAP1 knockdown (Supplementary Figures 5a and b). Furthermore, we did not observe any synergistic effect of AEG40730 with doxorubicin or etoposide on the viability of RH36 and RH41 cells (Supplementary Figures 5c–f), suggesting that SMC sensitisation of RMS cells is primarily executed via cIAP1 and the extrinsic cell death pathway. Taken together, these results suggest that SMCs could be used in combination with TNFα as a therapeutic approach to trigger the death of RMS cancer cells.

As a proof of principle, we next decided to test this approach in the human RMS cell line Kym-1 that has autocrine TNFα production39 and therefore should be sensitive to SMC treatment alone. Indeed, AEG40730 treatment of Kym-1 cells induced significant cytotoxicity (10-fold) and caspase activation (200-fold) when compared with a vehicle control (Supplementary Figures 6a–d). Furthermore, TNF receptor-1 (TNFR-1) knockdown (Supplementary Figure 2g) reduced cell death and caspase activation of Kym-1 cells in the presence of AEG40730, as compared with non-targeting siRNA-transfected cells (Supplementary Figures 6e–h). Similar data were obtained using the monomer SMC LCL161 that has better bioavailability in vivo (Supplementary Figure 7). These results demonstrate that cIAP1 depletion by SMCs sensitises Kym-1 cells to cell death in a TNFR-1-dependent manner.

SMC treatment inhibits the growth of Kym-1 RMS xenograft tumours

To test the efficacy of combined TNFα and SMC in a tumour model in vivo, we treated CD-1 female nude mice bearing established Kym-1 xenograft tumours with LCL161, or a vehicle control, two times a week for 2 weeks. LCL161 treatment significantly reduced the growth of Kym-1 tumours compared with vehicle-treated animals (Figure 6a). Concomitantly, the mean survival of LCL161-treated Kym-1 xenograft mice was significantly extended by 32 days compared with vehicle-treated mice (Figures 6b, P = 0.002). In a parallel experiment, we implanted Kym-1 cells and initiated SMC treatment the following day. We observed an even more striking effect of LCL161 on tumour growth and survival; LCL161-treated mice did not develop detectable tumours by 120 days postimplantation, whereas vehicle-treated animals developed sizeable tumours within 35 days (Figure 6c). Consequently, this treatment strategy led to the durable cure of LCL161-treated Kym-1 xenograft mice (Figures 6d, P = 0.002). Western blot analysis of excised tumours from Kym-1 xenograft treated with LCL161 for 24 h confirmed cIAP1 depletion and an increase in cleaved poly (ADP-ribose) polymerase compared with vehicle-treated mice (Supplementary Figure 8), indicating that these tumour cells have undergone cell death. Immunohisto- chemistry analysis of these tumours also revealed a significant increase in cleaved caspase-3 in LCL161-treated mice compared with vehicle (Figure 6e). IGF2BP1 protein levels were not affected by LCL161 treatment (Supplementary Figure 8).

Taken together, these results show that cIAP1 depletion by LCL161 inhibits the growth of Kym-1 tumours by sensitising them to TNFα-induced cell death, thus significantly increasing the survival of Kym-1 tumour-bearing mice. These results attest to the potential of using SMCs for the treatment of RMS tumours.

DISCUSSION

cIAP1 is a key regulator of apoptosis and cancer cell survival by controlling the NF-κB signalling and extrinsic cell death pathways.5 In an attempt to better understand the mechanism of translational regulation of cIAP1 expression, we previously conducted RNA chromatography on the cIAP1 IRES and identified four potential ITAFs binding specifically to it, including IGF2BP1.

Here, we found that IGF2BP1 is overexpressed in a panel of human RMS cell lines and in human primary RMS tumours when compared with skeletal muscle or myoblasts (Figures 1a and d and Supplementary Figures 1a–d). These findings are consistent with the classification of IGF2BP1 as an oncofetal protein that is re- expressed in a variety of cancers (reviewed in Yisraeli,13 Bell et al.22 and Yaniv and Yisraeli40). Interestingly, IGF2BP1 mRNA levels are elevated in all RMS cell lines analysed compared with HSMM (Supplementary Figure 1e) and may explain in part the observed increase in protein. Currently, IGF2BP1 expression regulation is not well understood;22 hence, it would be interesting to investigate its expression pattern during human embryogenesis and carcinogen- esis. cIAP1 showed the same pattern of expression as IGF2BP1 (Figures 1a and b), although cIAP1 mRNA levels were significantly decreased in RMS cell lines compared with HSMM (Supplementary Figure 1e). This observation, along with the fact that IGF2BP1 was identified as a cIAP1 IRES-binding protein, led us to hypothesise that IGF2BP1 regulates cIAP1 IRES-mediated translation. Indeed, we confirmed that IGF2BP1 is a bona fide ITAF that drives cIAP1 protein expression by enhancing cIAP1 translation specifically through binding and modulation of its IRES activity (Figures 1–3). The mechanisms by which ITAFs modulate IRES-mediated translation are still not fully understood; it is possible that IGF2BP1 interacts with other known cIAP1 ITAFs, such as NF458,41 and p97.6,42 Interestingly, IGF2BP1 was shown to enhance HCV IRES- mediated translation via the 3′-UTR,20 and recent reports point to a role of IGF2BP1/2 in mediating IGF2 IRES translation upon mammalian target of rapamycin activation.19,43 IGF2BP1 is thus emerging as a critical ITAF that regulates the synthesis of viral and cellular proteins.

IGF2BP1 has also been emerging as an important oncogenic factor;22 however, its role in facilitating the apoptotic resistance of cancer cells has not been previously determined. We found that reducing IGF2BP1 levels sensitises RMS cells to TNFα-mediated cell death in a cIAP1-dependent way (Figure 4). Given the importance of cIAP1 in sensitising RMS cells to TNFα-induced cell death, we used a class of pharmacologics that target the IAPs, Smac mimetics.44 SMCs have been shown to be effective in the treatment of several cancers and are currently in phase I and II clinical trials (reviewed in Gyrd-Hansen and Meier5 and LaCasse et al. 45). As expected, SMCs sensitised RH36 and RH41 RMS cells to cell death, in the presence of TNFα or TRAIL, similar to IGF2BP1 downregulation (Figure 5 and Supplementary Figures 3 and 4). We also observed the same sensitisation of Kym-1 cells to cell death in the presence of the SMCs AEG40730 and LCL161 without the addition of exogenous TNFα, in a TNFR1-dependent manner (Supplementary Figures 6 and 7). Indeed, Kym-1 cells were previously shown to have autocrine TNFα production and to mediate cell death via TNFR1.39,46 Importantly, our findings are corroborated by a recent paper showing that SMCs synergise with Lexatumumab (a TRAIL receptor2 agonist antibody) to induce cell death in a variety of RMS cells in an RIP1-dependent manner.37

Finally, we tested the synergistic effects of SMC and TNFα on the growth of RMS tumours in vivo. LCL161 significantly decreased the growth of established Kym-1 xenograft tumours and conse- quently extended the mean survival of mice by 32 days compared with vehicle-treated mice (Figure 6b). Importantly, LCL161 treatment was able to prevent the onset of tumourigenesis when SMC treatment commenced one day postimplantation (Figure 6c). Taken together, these results attest to the efficacy of SMC in reducing Kym-1 tumour burden and to the potential of using this therapeutic approach for the treatment of rabdomyosarcoma, especially if the tumours are detected early. Our observations are of relevance since a recent report on the initial testing of LCL161 by the Paediatric Preclinical Testing Program did not find any significant delay in the growth of several RMS xenograft tumours
when used as a single agent.47 Although therapies involving TNFα or TRAIL3 may not be a viable option because of toxicity of these agents, our results suggest that therapeutic strategies that target IAPs in combination with other TNFR signalling pathway-inducing agents may be effective in the treatment of RMS, as we have recently shown with oncolytic viruses.48

In summary, we identified IGF2BP1 as a critical regulator of cIAP1 expression and cell death resistance in RMS. Our results not only strongly argue for combined use of SMCs and TNFR- stimulating agents as a potential therapeutic approach for RMS, they also hint at the potential use of IGF2BP1 as a biomarker to identify which tumours would be responsive to such a combina- torial therapy.

MATERIALS AND METHODS

Cell culture, reagents, expression constructs and transfection Frozen and formalin-fixed, paraffin-embedded human paediatric RMS tumours and normal skeletal muscle samples were obtained from the Children’s Hospital of Eastern Ontario Department of Pathology after institutional ethics board approval. Control human and fetal skeletal muscle whole lysates were purchased from Novus Biologicals (Oakville, ON, Canada). Human RMS cell lines (RH18, RH30, RH36, RD and RH41) were a generous gift from Dr P Houghton (Department of Hematology-Oncology, St Jude Children’s Research Hospital, Memphis, TN, USA) and were cultured in RMPI-1640 media. The human RMS cell line Kym-1 was purchased from the JCRB (Osaka, Japan) and cultured in Dulbecco’s modified Eagle’s medium-F12. Primary HSMMs were purchased from Lonza (Mississauga, ON, Canada) and HSMM total RNA was purchased from Amsbio (Lake Forest, CA, USA). HEK293 cells were purchased from the ATCC (Manassas, VA, USA) and cultured in Dulbecco’s modified Eagle’s medium. HEK293 cells stably expressing a non-targeting or IGF2BP1-shRNA (Supplementary Table 1) from a tetracycline-inducible promoter (pTRIPz plasmid) were maintained in TET-free Dulbecco’s modified Eagle’s medium supplemented with 2 μg/ml puromycin. Transient siRNA transfections (Supplementary Table 1) were performed using Lipofectamine RNAiMax reagent (Invitrogen, Burlington, ON, Canada). Transient DNA transfections were performed using JetPRIME reagent (Polypus Transfection, Illkirch, France). The GFP-IGF2BP1 plasmid was a gift from Dr Stefan Hüttelmaier (Institute of Molecular Medicine, Martin Luther University of Halle, Halle, Germany); the pTurboGFP-dest1 (Evrogen, Moscow, Russia) was used as a transfection control.

Western blot analysis

Western blot analysis were performed as described previously.41 Membranes were probed with antibodies against IGF2BP1 (IMP1 clone D9; Santa Cruz Biotechnologies, Dallas, TX, USA), cIAP1 (RIAP1; Graber et al.8), XIAP (RIAP3; Baird et al.10 and Holcik et al.49), glyceraldehyde 3- phosphate dehydrogenase (GAPDH) (BD Biosciences, Mississauga, ON, Canada), cleaved poly (ADP-ribose) polymerase (clone D64E10; Cell Signalling, Beverly, MA, USA) and cleaved caspase-3 (clone 5A1E; Cell Signalling).

5′-UTR and IRES reporter assays

The monocistronic and bicistronic cIAP1 5′-UTR reporter constructs were previously described and validated.6,8,10 Relative translation activity of cIAP1 5′-UTR was assayed using monocistronic plasmids as described previously.8 Relative cIAP1 IRES activity was assayed using bicistronic
reporter plasmids as described previously.8

Polysome profiling and RT–qPCR analysis

HEK293 cells were transfected with 50 nM of non-targeting or IGF2BP1- targeting siRNA for 72 h and polysome profiling performed as described previously.41 cIAP1 and RPL13 A polysome-associated transcripts were quantified by RT–qPCR using specific primers (Supplementary Table 2). Steady-state mRNA levels were determined by RT–qPCR as described previously.41

Ultraviolet crosslinking RNA-binding assay

Ultraviolet-crosslinking RNA-binding assays were performed as described previously8 using [α-32P]UTP-labelled, in vitro-transcribed cIAP1 RNA probes (Supplementary Table 2) and recombinant GST-IGF2BP1.

Cell viability, cytotoxicity and caspase activity assays

A total of 1 × 104cells were transfected with siRNA for 48 h, treated with increasing concentrations of recombinant human TNFα or TRAIL (Enzo Life Sciences, Brockville, ON, Canada) and 48 h later assayed for cell respiration by Alamar Blue (Invitrogen). For cytotoxicity assays, siRNA-treated cells or cells treated with 100 nM of AEG40703036 or LCL16147,50 (kindly provided by Novartis Pharmaceuticals, Dorval, QC, Canada) or 0.1% dimethylsulph- oxide were incubated with 100 nM of YOYO-1 dye (Molecular Probes, Burlington, ON, Canada) in the presence of TNFα and incorporation monitored over 48 h using the INCUCYTE ZOOM Live-Cell Imaging System (Essen Bioscience, Ann Arbor, MI, USA). Caspase-3 and -7 activities were assayed using 1 μM of CellPlayer Caspase-3/7 reagent (Essen Bioscience) and monitored using the INCUCYTE ZOOM. For cIAP1 rescue experiments, RH36 cells were transfected with siRNA for 48 h, followed by transduction with LacZ (Adv-LacZ) or LacZ-cIAP1 (Adv-HIAP2) adenovirus before monitoring of cytotoxicity or cell death for 48 h. Fold cytotoxicity or cell death activity were calculated as the number of green fluorescence- positive cells divided by the total number of cells (confluence) at end point, compared with time zero.

Xenograft mouse model and Smac mimetic treatment Subcutaneous tumours were established by injecting 3 × 106 Kym-1 cells in Matrigel in the right flank of 6-week-old female CD-1 nude mice. For established tumours (~300–400 mm3), five mice were treated with vehicle (30% 0.1 M HCl, 70% 0.1 M NaOAc (pH 4.63)) or 50 mg/kg LCL16147,50 per os, two times a week for 3 weeks. In a parallel experiment, vehicle or LCL161 treatments commenced one day postimplantation for 2 weeks. Animals were killed when the tumour burden exceeded 2000 mm3. Tumour volume was calculated using (π)(W)2(L)/4, where W is the tumour width and L is the tumour length. All mice were maintained under barrier conditions and experiments were carried out according to protocols approved by the University of Ottawa Animal Care Facility.

Immunohistochemistry

Human RMS primary tumours and normal skeletal muscle formalin-fixed, paraffin-embedded sections were deparaffinised in xylene and rehydrated in 100–70% ethanol gradient. Heat-induced antigen retrieval was performed in 0.1 M EDTA and sections incubated with antibodies against IGF2BP1 (IMP1 clone D9) or cIAP1 (cIAP Pan, clone 315301; R&D Systems, Burlington, ON, Canada) for 1 h at 37 °C. DAB detection was performed using the Envision+anti-mouse-HRP Kit (Dako, Burlington, ON, Canada). The sections were analysed and scored for staining intensity, and the percentage of positive cells were estimated by a pathologist. For Kym-1 xenografts, established tumours treated with vehicle or 50 mg/kg LCL161 for 24 h were stained for cleaved caspase-3 (C92-605 antibody, BD Pharmingen) as described before.48