Loss of IGFBP7 Expression and Persistent AKT Activation Contribute to SMARCB1/Snf5-Mediated Tumorigenesis

ORIGINAL ARTICLE Loss of IGFBP7 expression and persistent AKT activation contribute to SMARCB1/Snf5-mediated tumorigenesis

J Darr, A Klochendler, S Isaac and A Eden

SMARCB1 (Snf5/Ini1/Baf47) is a potent tumor suppressor, the loss of which serves as the diagnostic feature in malignant rhabdoid tumors (MRT) and atypical teratoid/rhabdoid tumors (AT/RT), two highly aggressive forms of pediatric neoplasms. SMARCB1 is a core subunit of Swi/Snf chromatin remodeling complexes, and loss of SMARCB1 or other subunits of these complexes has been observed in a variety of tumor types. Here, we restore Smarcb1 expression in cells derived from Smarcb1-deficient tumors, which developed in Smarcb1 heterozygous p53 À / À mice. We find that while re-introduction of Smarcb1 does not induce growth arrest, it restores sensitivity to programmed cell death and completely abolishes the ability of the tumor cells to grow as xenografts. We describe persistent activation of AKT signaling in Smarcb1-deficient cells, which stems from PI3K (phosphatidylinositol 30-kinase)- mediated signaling and which contributes to the survival and proliferation of the tumor cells. We further demonstrate that inhibition of AKT is effective in preventing proliferation of Smarcb1-deficient cells in vitro and inhibits the development of xenografted tumors in vivo. Profiling Smarcb1-dependent gene expression, we find genes that require Smarcb1 and Swi/Snf for their expression to be enriched for extracellular matrix and cell adhesion functions. We find that Smarcb1 is required for transcriptional activation of Igfbp7, a member of the insulin-like growth factor-binding proteins family and a tumor suppressor in itself, and show that re-introduction of Igfbp7 alone can hinder tumor development. Our results define a novel mechanism for Smarcb1-mediated tumorigenesis and highlight potential therapeutic targets.

Oncogene (2014) 33, 3024–3032; doi:10.1038/onc.2013.261; published online 15 July 2013 Keywords: SMARCB1/Snf5; Swi/Snf; malignant rhabdoid tumors

INTRODUCTION necessary for p53- and possibly c-MYC-driven transcriptional 13,14 Mutation and inactivation of the tumor-suppressor gene SMARCB1 activation and cell-cycle control. Recently, aberrant activation (Snf5/Ini1/Baf47) have been well established as the underlying of the Hedgehog pathway in SMARCB1-deficient cells has been 15 mechanism leading to malignant rhabdoid tumors (MRT) and demonstrated as mediating tumorigenesis. atypical teratoid/rhabdoid tumors (AT/RT), two highly aggressive Further complicating our understanding of Smarcb1-associated forms of pediatric neoplasms.1–3 More recently, recurrent transformation is the finding that conditional knockout of Smarcb1 mutations in SMARCB1 in many other tumor types were also in primary cells results in rapid growth arrest and in p53-mediated 16 reported.4 SMARCB1 codes for BAF47, a core subunit of the Swi/Snf programmed cell death. This observation suggests that ATP-dependent chromatin remodeling complex that facilitates Smarcb1 transformation requires additional enabling conditions nucleosome re-positioning relative to the DNA sequence.5 such as mutations in additional genes or a unique cellular context Notably, mutations in SMARCB1 and in other subunits of the allowing for Smarcb1-associated transformation. Indeed, the Swi/Snf complex, such as the ATPases BRG1 and BRM, have been precise cellular origin of MRTs and AT/RTs is unknown. identified in a wide range of cancers, implicating the Swi/Snf Aspiring to further understand Smarcb1-associated transforma- complexes as a whole in tumor suppression.6,7 tion, we utilized our previously described MRT mouse model1 Swi/Snf complexes are generally regarded as transcriptional to establish two Smarcb1-deficient cell lines from tumors that co-activators/repressors. Accordingly, transformation associated developed spontaneously in Smarcb1 heterozygous, p53-deficient with SMARCB1 inactivation is considered to stem from transcrip- mice. Re-introduction of Smarcb1 to these murine tumor cells tional deregulation of target genes.4 Several mechanisms have resulted in complete loss of tumorigenicity but had only a minor been put forward when considering the tumor-suppressive effect on cell proliferation in vitro, suggesting that Smarcb1- properties of SMARCB1. SMARCB1 was shown to transcriptionally associated transformation may involve other mechanisms in regulate p16INK4a and/or p21, repress cyclin D1 and generally lead addition to the regulation of cell-cycle progression. This to inhibition of E2F activity.8–10 Consistent with these effects, cyclin possibility is further supported by a recent study that examined D1 deficiency inhibits tumor formation in Smarcb1 heterozygous SMARCB1 mutations in human familial schwannomatosis and mice.11 Moreover, gene expression profiling in SMARCB1-deficient found novel, cancer-causing mutations, which retain the ability to MRT tumors and in various cell lines following Smarcb1 conditional suppress cyclin D1.17 inactivation have also implicated SMARCB1 in the inhibition of Here we identify the AKT signaling pathway as a central axis in E2F-target genes.12,13 Further evidence suggests SMARCB1 as survival and proliferation of Smarcb1-deficient cells. We further

Department of Cell and Developmental Biology, The Alexander Silberman Institute of Life Sciences, The Hebrew University of Jerusalem, Jerusalem, Israel. Correspondence: Dr A Eden, Department of Cell and Developmental Biology, The Alexander Silberman Institute of Life Sciences, The Hebrew University of Jerusalem, Edmond J. Safra Campus, Jerusalem 91904, Israel. E-mail: [email protected] Received 6 October 2012; revised 10 May 2013; accepted 3 June 2013; published online 15 July 2013 AKT activation following SMARCB1 loss J Darr et al 3025 identify Igfbp7 as a Swi/Snf-regulated gene, the loss of which additional mechanisms as well. Moreover, these results reaffirm contributes to the tumorigenic capacity of Smarcb1-deficient cells. Smarcb1 as a potent tumor suppressor and assert the validity of this specific murine cellular system in studying transformation mediated by Smarcb1 deficiency. RESULTS Smarcb1-associated transformation relies on pathways additional to cell-cycle regulation Smarcb1-deficient tumor cells exhibit persistent AKT activation We established two cell lines from tumors that developed in and signaling Smarcb1 þ / À , p53 À / À mice.16 Both cell lines were Baf47 deficient Seeking to explain the Smarcb1-dependent differences in response due to spontaneous loss of heterozygosity of the Smarcb1 to serum withdrawal, and based on previous indications21–23 we wildtype allele (Figures 1a and b). Stable re-introduction of examined the levels of phosphorylated AKT (pAKT) in the two Smarcb1 into the tumor cells induced a drastic change in cell size tumor cell lines and found them to be significantly elevated in and morphology (Figure 1c), but unlike similar experiments in Smarcb1-deficient cells relative to their Smarcb1-proficient human tumor cells,12,18–20 Smarcb1 restoration had only a minor counterparts (Figure 2a). AKT phosphorylation levels depend on effect on growth rate in normal (10% fetal calf serum) serum external signals from serum components, and accordingly, in conditions (Figure 1d). Upon serum withdrawal (0.1% fetal calf Smarcb1-proficient cells lowering serum concentration abolished serum), Smarcb1-proficient tumor cells stopped proliferating and AKT phosphorylation. However, in Smarcb1-deficient cells, pAKT exhibited fast and dramatic cell death, an effect much delayed in was evident even in low serum. This result indicates that Smarcb1- Smarcb1-deficient tumor cells (Figures 1e–g). Both Smarcb1- deficient tumor cells have an aberrant and persistent activation of deficient and -proficient tumor cells failed to grow in soft agar AKT that is corrected when Smarcb1 is restored. (not shown), yet in vivo, when injected subcutaneously to AKT phosphorylation and activation promotes cell growth and NOD-SCID mice, tumors developed exclusively from Smarcb1- proliferation while inhibiting apoptosis, an effect which is mediated deficient cells (Figure 1h). The fact that following restoration through phosphorylation of various downstream targets. One such of Smarcb1, murine MRT cells continue to proliferate but lose target is the transcription factor FoxO1, whose phosphorylation tumorigenic capabilities indicate that Smarcb1 promotes cell prevents it from entering the nucleus and promoting expression transformation not only through cell-cycle regulation but via of pro-apoptotic genes and cyclin-dependent kinase inhibitors.24–28

Figure 1. Re-expression of Smarcb1 in Smarcb1-deficient tumor cells affects cell morphology, cell proliferation and cell survival. (a) Two cell lines (167, 365) were established from tumors that spontaneously developed in Smarcb1 heterozygous, p53-deficient mice. PCR analysis indicates loss of the wild-type Smarcb1 allele in tumor cells (Tu) when compared with tail DNA (Tl). (b) Western blot depicting Smarcb1 stable re-expression in two Smarcb1-deficient tumor cell lines: 167 and 365. Control cells were infected with control vector expressing green fluorescent protein (pMIG). (c) Phase-contrast images depicting characteristic morphological changes following stable re-expression of Smarcb1 in Smarcb1-deficient tumor cell lines. (d) Growth curve of Smarcb1-deficient and -proficient cells cultured in 10% serum over a period of 4 days. Error bars indicate s.d. for three repeats. No statistical significance was found when comparing the growth rate of Smarcb1-proficient cells to their deficient isogenic lines (T-test, P-value40.05). (e) Proliferation and apoptosis of Smarcb1-deficient and -proficient tumor cells following 24 h of serum depletion (0.1% fetal bovine serum) as estimated by BrdU incorporation and sub-G1 fraction respectively. Blue—DAPI, red—BrdU. (f) Quantification of proliferating cells, as estimated by BrdU incorporation, in cells depleted of serum (0.1%) for 24 h. Error bars represent s.d. of three repeats, n ¼ 300 nuclei. *Indicated P-value for T-test. (g) Quantification of apoptosis, as estimated by sub-G1 fraction, following 24 h of serum depletion (0.1%). (h) NOD-SCID mice were xenografted with Smarcb1-deficient tumor cells complemented with vector expressing Smarcb1 or control vector. Only cells lacking Smarcb1 developed into tumors (n ¼ 10 per cell line).

& 2014 Macmillan Publishers Limited Oncogene (2014) 3024 – 3032 AKT activation following SMARCB1 loss J Darr et al 3026

Figure 2. Aberrant AKT signaling in Smarcb1-deficient cells. (a) Western blot showing elevated pAKT levels in Smarcb1-deficient tumor cell lines relative to Smarcb1-proficient cell lines. This difference is most evident under low serum conditions. Downstream to AKT, Smarcb1-deficient tumor cells have reduced levels of p27 and high S6 ribosomal protein phosphorylation, particularly under serum starvation. (b) Real-time reverse transcriptase–PCR analysis showing Smarcb1-dependent transcriptional induction of two FoxO1 targets, p27 and Bim, under low serum (0.1%). *P-valueo0.01, **P-valueo0.001 (c) Western blot showing effective inhibition of AKT phosphorylation using increasing concentrations of the AKTi 1/2 inhibitor. (d) Quantification of the effect of AKT inhibition on cellular proliferation using a BrdU incorporation assay. Cells were grown in medium supplemented with 10% serum and 10 mM AKTi or dimethyl sulfoxide (DMSO) as control for 24 h. Error bars represent s.d. of three repeats, n ¼ 300 nuclei, blind procedure. *P-valueo0.05, **P-value o0.01. Note that AKTi inhibits proliferation of Smarcb1-deficient but not -proficient cells. (e) Western blot of tumor lysate depicting in vivo inhibition of AKT phosphorylation 9 h following a single dose of 1 mg/g of the AKT inhibitor Triciribine. (f) Xenografted Smarcb1-deficient tumor progression in mice treated with a daily dose of 1 mg/g of the AKT inhibitor Triciribine relative to controls. Error bars represent s.e.m. for six tumors overall for each group. *P-valueo0.05 as estimated by T-test.

In both tumor cell lines, re-introduction of Smarcb1 resulted in in Smarcb1-deficient cells depends on PI3K activity, we used the reduction of AKT phosphorylation and its downstream signaling, as PI3K inhibitor LY294002. As before, reducing serum concentration evident by the transcriptional induction of the cyclin-dependent resulted in reduced AKT phosphorylation in both Smarcb1- kinase inhibitor p27 and the pro-apoptotic gene Bim, two deficient and -proficient cells. However, while in Smarcb1- transcriptional targets of FoxO1, and by the decreased expressing cells AKT phosphorylation was completely abolished phosphorylation of S6 ribosomal protein (Figures 2a and b). These in low serum, it persisted in Smarcb1-deficient cells (Figures 2a results demonstrate that Smarcb1 deficiency in these cells leads to and 3a). This persistent, Smarcb1-dependent AKT phosphorylation activation of AKT signaling and provide a mechanistic explanation was suppressed following treatment with LY294002, suggesting to the increased cell proliferation and survival of Smarcb1-deficient that it is mediated by PI3K activation (Figure 3a). cells, particularly in low serum, as observed in Figure 1. PI3K-mediated AKT phosphorylation and activation is stimu- lated upon diverse signals, including growth factor receptors, 27,29 AKT inhibition hinders tumor cell proliferation insulin, integrins, cytokines and so on. Indeed, a western blot for phospho-Tyrosine (pTyr) in Smarcb1-deficient and -proficient To further validate the role of AKT signaling in survival and cells reveals multiple differentially phosphorylated proteins proliferation of Smarcb1-deficient cells and explore the possible (Figure 3b), suggesting that several players may contribute to therapeutic potential in targeting this pathway, we assayed the AKT activation. sensitivity of the tumor cells to an AKT inhibitor. Inhibition of AKT activity with an AKT1/2-specific inhibitor reduced proliferation rates of Smarcb1-deficient cells to a level comparable to that of Transcriptional response to Smarcb1 restoration their Smarcb1-proficient counterparts, while having little, if any, Our findings thus far provide evidence for aberrant activation effect on Smarcb1-proficient cells (Figures 2c and d). To assess the of AKT in Smarcb1-deficient tumor cells, which is mediated contribution of AKT activation to survival of Smarcb1-deficient by PI3K and contributes to transformation. However, it remained tumor cells in vivo, we treated mice carrying xenografted Smarcb1- unclear how deficiency in Smarcb1, a member of Swi/Snf deficient tumors with the AKT inhibitor Triciribine and found it to chromatin remodeling complexes, can cause such an effect. inhibit tumor growth (Figures 2e and f). Overall, these results Swi/Snf complexes are recognized mostly as transcriptional indicate that the AKT signaling pathway contributes to survival of co-activators and co-repressors, and previous findings show Smarcb1-deficient tumor cells and suggest that inhibition of AKT that Smarcb1-dependent transformation is mediated by its may be a beneficial approach for suppression of Smarcb1-deficient transcriptional targets and not by promoting mutagenesis.11,30 tumors, such as MRT and AT/RT. We therefore used DNA microarrays to characterize Smarcb1- dependent transcriptional changes in tumor cells. To enrich Inhibition of phosphatidylinositol 30-kinase (PI3K) abolishes for transcriptional targets that are directly regulated by aberrant AKT activation in Smarcb1-deficient tumor cells Smarcb1, we profiled gene expression in both the tumor cell Signaling leading to AKT phosphorylation is often mediated by lines following Smarcb1 re-introduction, as soon as the BAF47 PI3K.27 To determine whether the elevated level of pAKT observed protein could be detected. Following Smarcb1 re-expression,

Oncogene (2014) 3024 – 3032 & 2014 Macmillan Publishers Limited AKT activation following SMARCB1 loss J Darr et al 3027 Table 1. Gene ontology terms enriched in gene expression changes following Smarcb1 re-expression

Name 167 365

NES FDR NES FDR

Upregulated in Smarcb1 þ Cell adhesion 1.888 o1e-3 1.909 o1e-3 Receptor activity 1.668 o1e-3 1.481 o1e-3 Plasma membrane 1.490 o1e-3 1.508 o1e-3 Cell migration 1.750 o1e-3 1.647 0.001 Actin cytoskeleton 1.655 o1e-3 1.603 0.004 EMT 2.103 o1e-3 1.591 0.016 Extracellular space 2.223 o1e-3 1.389 0.017 Cell-cell adhesion 1.582 o1e-3 1.490 0.028 Integrin signaling pathway 1.650 0.006 1.627 0.004

Downregulated in Smarcb1 þ DNA repair À 1.826 o1e-3 À 1.479 o1e-3 Cell cycle À 1.943 o1e-3 À 1.610 o1e-3 DNA replication À 2.245 o1e-3 À 2.029 o1e-3 DNA damage signaling À 1.943 o1e-3 À 1.620 0.008 Abbreviations: EMT, epithelial–mesenchymal transition; FDR, false discovery rate; NES; Normalized Enrichment Score. Gene set enrichment analysis was used to study the transcriptional response to Smarcb1 re-expression in Smarcb1-deficient tumor cell lines. Listed are Gene ontology categories found to be enriched among genes upregulated or downredugated Figure 3. AKT phosphorylation in Smarcb1-deficient tumor cells is following expression of Smarcb1 in both the tumor cell lines. mediated through PI3K activity. (a) Western blot depicting inhibition of Smarcb1-dependent AKT phosphorylation in Smarcb1-deficient tumor cells treated with 20 mM of the PI3K inhibitor LY20029 for 24 h relative to dimethyl sulfoxide (DMSO)-treated control cells. In low serum, AKT remains phosphorylated only in Smarcb1-deficient cells. Table 2. Transcription factor-binding motifs found to be enriched in This Smarcb1-dependent AKT phosphorylation is eliminated by promoters of upregulated or downredugated genes following re- LY20029 treatment. (b) Western blot for phospho-Tyrosine residues expression of Smarcb1 (pTyr) in Smarcb1-deficient and -proficient tumor cells cultured under serum depletion or following 10% serum induction for the Motif Name 167 365 indicated times. Arrows mark multiple bands unique to Smarcb1- deficient cells. ON, over night. NES FDR NES FDR

Upregulated in Smarcb1 þ 490 and 371 genes were upregulated at least two fold in cell CCAWATAWGGMNMNG SRF 2.18 o1e-3 1.49 0.01 GNNTTGTTTACNTT FOXO1 1.59 o1e-3 1.47 0.01 lines 167 and 365, respectively, with an overlap of 170 genes GKSRKKCAGMCANCY SMAD4 1.50 0.01 1.37 0.02 (Supplementary Figure S1a). In all, 186 and 209 genes were downregulated at least two fold in cell lines 167 and 365, Downregulated in Smarcb1 þ SCGGAAGY ELK1 À 1.39 o1e-3 À 1.18 o1e-3 respectively, with only 21 genes in common (Supplementary SGCGSSAAA E2F1 À 1.68 o1e-3 À 1.46 0.01 Figure S1a). NNANCACGTGNTNN MAX À 1.48 o1e-3 À 1.24 0.04 31,32 Gene set enrichment analysis (GSEA) showed that genes Abbreviations: FDR, false discovery rate; NES; Normalized Enrichment that are upregulated following re-introduction of Smarcb1 are Score. Gene set enrichment analysis was used to identify putative enriched for such gene ontology categories as cell adhesion, transcription factor-binding sites, which are over-represented in promoters migration and extracellular components (Table 1, false discovery of genes which responded to re-expression of Smarcb1 in tumor cell lines. rate o0.001) and include central players such as fibronectin, paxilin, integrins and MMPs. These transcriptional changes offer several possible mechanistic explanations to the observed morphological changes in cell shape and size following Smarcb1 similarity (as defined by the GSEA ‘leading edge’ group) include re-introduction. Consistent with similar studies in human MRT extracellular matrix (ECM) components such as Fibronectin1 and cells,13 genes downregulated following re-introduction of Smarcb1 Integrinb3 and other secreted molecules (Supplementary Table S1) in our experiment were found to be enriched for categories and exhibit as a whole statistically significant enrichment for gene associated with DNA repair and replication and cell-cycle ontology terms related to ECM and adhesion (Supplementary regulation (Table 1). As Swi/Snf complexes are known to be Table S2). As Baf47 (Smarcb1) is a core component of Swi/Snf recruited by transcription factors, we used GSEA to search for complexes, the similarity in transcriptional response to re- enrichment in common DNA-binding motifs. Among the genes introduction of different complex subunits indicates that at least downregulated by Smarcb1, we found significant enrichment for some of the transcriptional perturbations in Smarcb1-deficient targets of E2F1, Max and Elk1, while for genes upregulated by cells reflect loss of Swi/Snf complex functionality in the absence of Baf47. However, it remains unknown to what degree these Smarcb1 we found a significant enrichment for targets of FoxO1, 35 SRF and Smad4 (Table 2). changes represent similarity in transformation mechanism. There is a significant correlation between the transcriptional changes observed in our study and those following re-expression Inhibition of xenografted tumor growth following expression of of other subunits of the Swi/Snf complex in other experimental Igfbp7 in Smarcb1-deficient tumor cells systems, specifically Smarca4 (BRG1) and Smarce1 (Baf57)33,34 Expression data also revealed several candidate genes whose low (Supplementary Figure S1b). The genes which contribute to this expression in Smarcb1-deficient cells may explain Smarcb1-

& 2014 Macmillan Publishers Limited Oncogene (2014) 3024 – 3032 AKT activation following SMARCB1 loss J Darr et al 3028

Figure 4. Igfbp7 transcriptional repression in Smarcb1-deficient tumor cells contributes to transformation. (a): Real-time reverse transcriptase– PCR verifying Igfbp7 and Cav1 induction following Smarcb1 re-expression in tumor cells. Fold induction is normalized to b-actin. (b) Western blot depicting Igfbp7 and Cav1 protein in Smarcb1-proficient cells. (c) Xenografted tumor growth curve depicting suppression of growth following Igfbp7 expression in Smarcb1-deficient tumor cells. Average tumor volume in mm3 is plotted (n ¼ 6). From day 9 onwards, the difference between IGFBP7 expressing cells vs control tumor cells becomes statistically significant (P-valueo0.05, T-test). (d) Xenografted tumor growth following Cav1 expression in Smarcb1-deficient tumor cells (167 cell line, n ¼ 6). (e) Representative images from immunohistochemical staining of xenografted tumor sections. In tumors derived from line 167, introduction of IGFBP7 results in reduced phosphorylation of AKT and S6 ribosomal protein. In line 365, only reduction in phosphorylation of S6 is evident. (f) Western blot showing pAKT and pS6 in xenograft tumor lysate. Quantification is consistent with phosphorylation levels as apparent in panel e.(g) Quantification of PCNA-positive cells using immunohistochemical staining reveals a reduction in cell proliferation in Igfbp7-expressing xenografts. *Indicated P- value for t-test.

associated transformation and the increased levels of AKT marked reduction in proliferating cell nuclear antigen (PCNA)- phosphorylation. We considered two such candidates which were positive cells, indicating reduced in vivo proliferation of Igfbp7- among the most upregulated genes following Smarcb1 restora- expressing tumor cells (Figures 4e–g). These results indicate that tion: Caveolin1, which is involved in modulating epidermal growth the loss of Igfbp7 expression in Smarcb1-deficient tumor cells factor receptor or transforming growth factor-b activity,36,37 and contributes to the transformed phenotype, yet it is still unclear IGFBP7, which is related to the insulin-like growth factor binding whether Igfbp7 acts by interfering with IGFR signaling. proteins family, whose members have been demonstrated to bind Igfbp7 has been studied as a modulator of angiogenesis and and sequester IGF-I and thus attenuate signaling through the IGF1 was found to inhibit tumor angiogenesis in several contexts receptor.38–41 Both genes were induced greatly following Smarcb1 and proposed to block vascular endothelial growth factor re-expression, as validated by quantitative reverse transcriptase– receptor.42–45 We used the endothelial marker CD31 to examine PCR and by western blot (Figures 4a and b). the vasculature of Igfbp7-expressing xenograft tumors and control To test whether these proteins may contribute to AKT counterparts. Consistent with previous publications, expression of phosphorylation and to Smarcb1-dependant tumorigenicity, we Igfbp7 in tumor cells had a marked effect on distribution and size expressed them in Smarcb1-deficient tumor cells (Supplementary of blood vessels (Supplementary Figure S3). It is therefore possible Figure S2). In culture, re-expression of Igfbp7 or Cav1 did not lower that the growth-suppressive effect of Igfbp7, observed only the levels of pAKT or pS6 and did not confer any measurable in vivo, stems from an effect on angiogenesis. Although expression effect on cell morphology or proliferation (data not shown). of Igfbp7 alone did reduce tumor growth, it did not completely However, expression of Igfbp7 (but not Cav1) inhibited growth of abolish tumor development and did not correct the persistent xenografted tumor cells (Figures 4c and d). As expression of Igfbp7 activation of AKT. We therefore conclude that transformation in tumor cells inhibited tumor growth but did not prevent it caused by loss of Smarcb1 involves components additional to completely, we could examine its effect on the AKT signaling Igfbp7 transcriptional downregulation. pathway in vivo. In tumor cell line 167, we found that Igfbp7 expressing tumors had reduced levels of pAKT compared with control 167 cells (Figures 4e and f). This effect was not evident in DISCUSSION line 365. However, in both the cell lines, expression of Igfbp7 In this study, we took advantage of a previously described mouse resulted in significantly reduced S6 ribosomal protein phosphor- model for Smarcb1/Snf5/Ini1/Baf47-deficient tumors.16 Focusing ylation (Figures 4e and f). Consistent with these results, we find a on the contribution of Smarcb1 deficiency to the etiology of these

Oncogene (2014) 3024 – 3032 & 2014 Macmillan Publishers Limited AKT activation following SMARCB1 loss J Darr et al 3029 tumors, we study phenotypic and molecular consequences of proliferation of Smarcb1-deficient tumor cells and demonstrate Smarcb1 re-introduction into tumor cells. As opposed to previous that AKT aberrant activation is dependent on absence of Smarcb1 studies in human cells,12,18–20 re-introduction of Smarcb1 in this (Figure 3). system did not result in growth arrest (Figure 1d) but nevertheless One the most upregulated genes following Smarcb1 re- abolished the tumorigenic capacity of the cells (Figure 1h). This introduction to tumor cells was Igfbp7, which was recently outcome may reflect a difference between human and mouse highlighted as a potential tumor-suppressor gene in various solid Smarcb1-deficient tumors; however, p53 deficiency may also tumors through a mechanism which remains to be eluci- contribute to the escape from growth arrest following Smarcb1 re- dated.43,44,51–53 Other members of the IGFBP family were found expression. Even though the tumor cells are p53 deficient, they to bind IGF and affect its bioavailability44,54 but IGFBP7 is less completely lose tumorigenic capability following re-introduction related to this group and is thought to have only low affinity to of Smarcb1. Here we focus on understanding this critical Smarcb1- IGF. Interestingly, IGFBP7 was shown to have a high affinity to dependent component of transformation. insulin,55 which was suggested to act as an autocrine growth factor in Rhabdoid tumor cells.50 Re-expression of Igfbp7 alone in Smarcb1-deficient tumor cells Persistent AKT activation in Smarcb1-deficient tumor cells inhibited xenograft growth in vivo and reduced AKT pathway Smarcb1-deficient tumor cells exhibit resistance to apoptosis and activity, yet Igfbp7 deficiency does not fully account for Smarcb1- increased proliferation, particularly in low serum conditions, as associated transformation: while re-introduction of Smarcb1 compared with Smarcb1-expressing cells (Figure 1). At the completely prevented the development of xenograft, Igfbp7 had molecular level, we find persistent, PI3K-dependent, activation of only an inhibitory effect (Figures 1h and 4c). In addition, AKT in serum-deprived Smarcb1-deficient cells (Figure 2) that is expression of IGFBP7 alone did not reduce AKT phosphorylation abolished upon re-expression of Smarcb1. Accordingly, in in cultured tumor cells. IGFBP7 is a secreted factor and is known to Smarcb1-deficient cells we identify key downstream effects of depend on ECM components for its activity.56 ECM and adhesion- AKT activation, including increased S6 phosphorylation and associated functions are underexpressed in Smarcb1-deficient cells reduced p27 and Bim expression, which are known to promote (including fibronectin, integrins and MMPs, which were shown to proliferation and prevent apoptosis (Figures 2a and b). The function in coordination with IGFBPs56–58). It is thus possible that dependence of tumor cells on AKT activity for survival and the tumorigenic effect of Smarcb1 deficiency is the result of mis- proliferation is also evident from the effect of AKT inhibition on regulation of multiple genes, which together lead to AKT cell proliferation, which is specific to Smarcb1-deficient cells phosphorylation and result in transformation. Further support to (Figure 2d). The ability of an AKT inhibitor to attenuate growth of this possibility comes from the western blot for pTyr, where re- xenografted Smarcb1-deficient tumor cells further demonstrates introduction of Smarcb1 appeared to attenuate the the relevance of AKT signaling in the progression of these tumors phosphorylation levels of multiple proteins (Figure 3b). (Figure 2f). In conclusion, our results demonstrate that targeting AKT signaling pathway may prove effective in treatment of Smarcb1- Smarcb1-dependent transcription in tumor cells deficient tumors. As more tumor types are found to harbor 59 To further investigate the molecular basis of cellular phenotypes mutation in Smarcb1 or to lose its expression, our results may be brought about by Smarcb1 deficiency, we compared gene applicable for such tumors as well and not only to MRT and AT/RT. expression profiles of Smarcb1-proficient and -deficient tumor Our results further show that expression of Igfbp7 is dependent cells. The similarity between Smarcb1-dependent transcription and on Smarcb1/Swi/Snf activity and that Igfbp7 has a tumor- BRG1- or BAF57-dependent transcription indicates that at least suppressing effect in SMARCB1-deficient tumor cells. Igfbp7 was part of the transcriptional alterations in Smarcb1-deficient cells are not studied in human rhabdoid tumors; however, comparison of the result of loss of Swi/Snf complex functionality in the absence Igfbp7 expression across a spectrum of human tumor types and of Smarcb1. However, it remains unknown to what degree these tumor cell lines reveals that Igfbp7 expression levels are among changes reflect a similar mechanism of transformation.35 the lowest in human MRT and AT/RT samples (Supplementary Our analysis identifies candidate transcription factors that may Figure S4). This suggests that although its mechanism of action recruit Swi/Snf complexes as co-activators/repressors in tumor remains to be explored, IGFBP7 repression may also have a role in cells. We find that Smarcb1 responsive genes are enriched in promoting human tumorigenesis following loss of SMARCB1. binding sites for E2F1, Elk1, SRF and Smad4, which have been reported previously to interact with Swi/Snf complex members for MATERIALS AND METHODS their transcriptional activity.46–48 The GSEA analysis also defines the list of putative target genes, which have the relevant binding Cell line establishment and culture site for these transcription factors, and which also change in Spontaneous tumors developing in Scmarb1 heterozygote p53-deficient expression soon after BAF47 re-expression. It remains to be mice were removed, dissected and suspended in serum-free DMEM (Dulbecco’s modified Eagle’s medium) containing 0.75 g/ml collagenase determined which of the responsive genes are indeed directly (Sigma, Rehovot, Israel, No. c5138) for 2 h. Following over night incubation, regulated by Swi/Snf complexes. cells were mechanically separated by pipetting and suspended in growth medium. All cells were grown in DMEM supplemented with 10% Hyclone Transcriptional downregulation of Igfbp7 in Smarcb1-deficient fetal bovine serum, penicillin (50 mg/ml), streptomycin (50 mg/ml), 2 mM tumor cells contributes to their transformed phenotype l-Glutamine, 0.1 nM non essential amino acids, 0.1 mM b-mercaptoethanol and 1 mM sodium pyruvate. For low serum conditions, cells were washed Although multiple mechanisms can lead to AKT activation, we twice in phosphate-buffered saline (PBS) before being transferred to found that inhibition of PI3K can attenuate the persistent medium containing 0.1% fetal bovine serum. PI3K inhibitor LY20094 phosphorylation of AKT in Smarcb1-deficient cells (Figure 3), (Sigma), AKT inhibitor 1/2 (Calbiochem, Darmstadt, Germany, AKT inhibitor implicating signaling upstream of AKT in its persistent activation. VIII No. 124018) were solubilized according to the manufacturer’s Activation of the IGF1R-AKT signaling pathway was previously instructions before addition to medium in indicated concentrations. observed in human MRT and AT/RT tumor sections, and studies in cell lines derived from human Rhabdoid tumors suggest aberrant Xenografts and mice handling activity of the IGF1-AKT signaling pathway and dependence on Experiments were approved by the Israeli Animal Care and Use Committee 21–23,49,50 this pathway for tumor cell survival. Our results (NS-09-12222-3). Mice were kept in specific pathogen-free approved emphasize the central role this pathway has in survival and facility. For xenografting, weaned Nod-SCID males aged 4 weeks were

& 2014 Macmillan Publishers Limited Oncogene (2014) 3024 – 3032 AKT activation following SMARCB1 loss J Darr et al 3030 sedated with Isoflurane. Using a 27-G needle one million cells suspended manufacturer’s instructions. All other procedures were performed using in a total volume of 100 ml growing media were injected subcutaneously to standard techniques and kits as described in Barzily-Rokni et al.63 each flank. When treated with the AKT inhibitor Triciribine (Merck, Darmstadt, Germany, AKT inhibitor V No. 124012), 1 mg per gram body Expression array and data analysis weight per day were injected intraperitoneally in a total volume of 100 ml Expression profiling was carried out on Affymetrix Mouse Gene 1.0st PBS. Inhibitor was injected daily starting the day following injection of cells, platform (Affymetrix, Santa Clara, CA, USA). GeneChip Whole Transcript for 10 days. Mice were weighed daily to monitor any changes in body (WT) Sense Target Labeling Assay (Affymetrix) was employed followed by mass, and their general health examined throughout the duration of the GeneChip Hybridization, Wash and Stain Kit (Affymetrix). Scanning was experiment. Tumors were measured using a caliper, and tumor volume performed with GeneChip Scanner 3000 7G (Affymetrix controlled by was calculated using the modified ellipsoidal formula.60 GeneChip Operating Software (Affymetrix). Raw probe intensities were normalized by the robust multiarray average method using the Affymetrix Retroviral vectors expression console program (Affymetrix). Probes with raw intensity value o5 were discarded. We applied GSEA analysis using the MSigDB Primers listed in Supplementary Information (Supplementary Table S3) collections31 to identify gene sets that were significantly enriched in were used to generate full-length Smarcb1 cDNA transcript by PCR either cell line. In our analysis, we considered gene sets with false discovery amplification of cDNA generated from human ES cells as listed below. The rate o0.01 to be significant. Expression data are available with GEO resulting amplicon was cloned into the BglII and BamHI sites of the pMig accession GSE46017. retroviral vector (Addgene, Cambridge, MA, USA, No. 9044) and sequenced. Igfbp7 and Cav1 were PCR amplified from cDNA generated from primary murine embryonic fibroblasts. Resulting amplicons were digested with Immunohistochemistry BamH1 and EcoR1/Sal1, respectively, and cloned into the pBabe-puro Tumors were harvested, fixed in 4% PFA-PBS and embedded in paraplast. retroviral vector. For generation of viral vectors, plasmids were co- Five micron sections were de-parafinized and rehydrated. Following 61 transfected with pCL-Eco into 293T cells using the calcium phosphate antigen retrieval in 10 mM citrate buffer pH 6.0 and inhibition of protocol. Infection was carried out for 2 sequential days with 8 mg/ml endogenous peroxidases with 3% H2O2, sections were blocked for Polybrene followed by selection with 5 mg/ml Puromycin for the pBabe- 30 min and then incubated with the primary antibody (anti-pAKT473 No. puro vector. 736e11 from Cell Signaling diluted 1:50 in TNB buffer, anti-P6S S240/S244 No. 2215s from Cell Signaling diluted 1:50 in CAS-Block, anti-Igfbp7 sc- 13095 from Santa Cruz diluted 1:50 in CAS-Block, anti-PCNA from Protein extraction and western blot analysis BioLegend, San Diego, CA, USA, No. 307901 diluted 1:50 in CAS-Block)

Proteins were extracted using p300 lysis buffer (20 mM NaH2PO4, 250 mM over night at 4 1C. Biotinylated secondary antibody diluted 1:200 was NaCl, 30 mM Na4P2O7*10H2O, 0.1% NP-40 and 5 mM EDTA) supplemented incubated for 30 min at room temperature. For pAKT staining, incubation with 1 mM dithiothreitol, 1 mM phenylmethanesulfonylfluoride, 1 mM pep- with Streptavidin-peroxidase conjugate from the PerkinElmer TSA kit statin, 1 mg/ml Aprotenin, 0.5 mg/ml Leupeptin and 1 mM Na3VO4. Following (Perkin-Elmer, Waltham, MA, USA) and the Biotynil Tyramide was carried 10 min on ice, the lysate was centrifuged at 12 000 r.p.m. and the pellet out in accordance to the manufacturer’s instructions. Sections were then discarded. The rabbit polyclonal anti-SNF5 was described previously,62 incubated with Avidin horseradish peroxidase (extraAvidin-Peroxidase anti-pAKT serine473 (Cell Signaling, Danvers, MA, USA, No. 4058), anti-t- E2886 Sigma) diluted 1:100 in 1% BSA-PBS for 15 min followed by AKT (Cell Signaling No. 11E7), anti-p27 (Santa Cruz, Dallas, TX, USA, sc-529), incubation with the DAB substrate from the DAB chromogen KIT DB801R anti-b-Actin (Abcam, Cambridge, MA, USA, ab6276), anti-p6S (Cell by Biocare Medical (Concord, CA, USA). Sections were stained with Signaling No. 2215s), anti-Igfbp7 (Santa Cruz sc-13095), anti-Cav1 (Cell hematoxylin and dehydrated before mounting. Signaling No. 3267x) and anti-pTyr (4G10 hybridoma) Secondary antibodies coupled to horseradish peroxidase (Jackson Immunoresearch Laboratories, West Grove, PA, USA). CONFLICT OF INTEREST The authors declare no conflict of interest. Growth curves, bromodeoxyuridine (BrdU) labeling and cell-cycle A standard MTT (3-[4,5-dimethylthiazol-2-yl]-2,5 diphenyl tetrazolium ACKNOWLEDGEMENTS bromide) assay was used for plotting growth curves: 1500 cells were We thank Dr Ittai Ben-Porath, Professor Eli Keshet and Professor Alex Levitzki for plated in triplets per day in a 96-well plate. For each time point, cells were reagents and helpful discussions. This work was funded by the Israel Cancer Research incubated for 1 h in 100 ml medium supplemented with a final concentra- Fund (ICRF, 2011-3094-PG) and by the Association for International Cancer Research tion of 400 mM MTT followed by two washes in PBS and extraction in 100 ml (AICR 03-109). dimethyl sulfoxide. Plate was read at 570 nm and the background absorbance at 690 nm subtracted. For BrdU labeling, cells were plated on coverslips, left to adhere over night and 10 mM BrdU was added for 30 min REFERENCES before fixation in ice-cold methanol for 10 min at room temperature. Fixed 1 Klochendler-Yeivin A, Fiette L, Barra J, Muchardt C, Babinet C, Yaniv M. The murine cells were treated with 2 M HCl for 30 min followed by two washes in NaBO4 SNF5/INI1 chromatin remodeling factor is essential for embryonic development 0.1 M pH 8.5 and two washes in PBS. For immunostaining, anti-BrdU (cell and tumor suppression. EMBO Rep 2000; 1: 500–506. proliferation kit, 1:300; GE Healthcare Life Sciences, Uppsala, Sweeden) was 2 Roberts CW, Leroux MM, Fleming MD, Orkin SH. Highly penetrant, rapid tumor- diluted in PBS supplemented with 0.5% BSA and 0.05% Tween for 45 min igenesis through conditional inversion of the tumor suppressor gene Snf5. Cancer followed by secondary antibody and DAPI (4,6-diamidino-2-phenylindole) Cell 2002; 2: 415–425. before mounting. Images were collected on a Nikon TE-2000 (Nikon, 3 Versteege I, Sevenet N, Lange J, Rousseau-Merck MF, Ambros P, Handgretinger R Melville, NY, USA) inverted microscope and processed using NIS-elements et al. Truncating mutations of hSNF5/INI1 in aggressive paediatric cancer. Nature software (Nikon). Identical camera and microscope settings were employed 1998; 394: 203–206. to allow valid comparison between images of Smarcb1-deficient and - 4 Wilson BG, Roberts CW. SWI/SNF nucleosome remodellers and cancer. Nat Rev proficient cells. For cell-cycle analysis, cells were trypsinized, washed and re- Cancer 2011; 11: 481–492. suspended in 300 ml PBS. Five milliliter ice-cold methanol was slowly added 5 Phelan ML, Sif S, Narlikar GJ, Kingston RE. Reconstitution of a core chromatin for fixation followed by incubation in À 20 1C for at least 1 h. Cells were remodeling complex from SWI/SNF subunits. Mol Cell 1999; 3: 247–253. subsequently washed, re-suspended in 1 ml PBS and incubated on ice for 6 Varela I, Tarpey P, Raine K, Huang D, Ong CK, Stephens P et al. Exome sequencing 30 min. Subsequently, cells were re-suspended in 1 ml propidium iodide identifies frequent mutation of the SWI/SNF complex gene PBRM1 in renal solution (PBS supplemented with 50 mg/ml propidium iodide and 250 mg/ml carcinoma. Nature 2011; 469: 539–542. RNase A ) and incubated in 37 1C for 30 min. 7 Reisman DN, Sciarrotta J, Wang W, Funkhouser WK, Weissman BE. Loss of BRG1/ BRM in human lung cancer cell lines and primary lung cancers: correlation with poor prognosis. Cancer Res 2003; 63: 560–566. RNA and DNA extraction, reverse transcription and real-time PCR 8 Medjkane S, Novikov E, Versteege I, Delattre O. The tumor suppressor hSNF5/INI1 Total RNA was extracted using the Qiagen RNeasy mini kit (Qiagen, modulates cell growth and actin cytoskeleton organization. Cancer Res 2004; 64: Valencia, CA, USA) for purification of total RNA and in accordance with the 3406–3413.

Oncogene (2014) 3024 – 3032 & 2014 Macmillan Publishers Limited AKT activation following SMARCB1 loss J Darr et al 3031 9 Oruetxebarria I, Venturini F, Kekarainen T, Houweling A, Zuijderduijn LM, 35 Wang X, Sansam CG, Thom CS, Metzger D, Evans JA, Nguyen PT et al. Oncogenesis Mohd-Sarip A et al. P16INK4a is required for hSNF5 chromatin remodeler-induced caused by loss of the SNF5 tumor suppressor is dependent on activity of BRG1, cellular senescence in malignant rhabdoid tumor cells. J Biol Chem 2004; 279: the ATPase of the SWI/SNF chromatin remodeling complex. Cancer Res 2009; 69: 3807–3816. 8094–8101. 10 Zhang ZK, Davies KP, Allen J, Zhu L, Pestell RG, Zagzag D et al. Cell cycle arrest 36 Aguilar RC, Wendland B. Endocytosis of membrane receptors: two pathways are and repression of cyclin D1 transcription by INI1/hSNF5. Mol Cell Biol 2002; 22: better than one. Proc Natl Acad Sci USA 2005; 102: 2679–2680. 5975–5988. 37 Le Roy C, Wrana JL. Clathrin- and non-clathrin-mediated endocytic regulation of 11 Tsikitis M, Zhang Z, Edelman W, Zagzag D, Kalpana GV. Genetic ablation of Cyclin cell signalling. Nat Rev Mol Cell Biol 2005; 6: 112–126. D1 abrogates genesis of rhabdoid tumors resulting from Ini1 loss. Proc Natl Acad 38 Oh Y, Nagalla SR, Yamanaka Y, Kim HS, Wilson E, Rosenfeld RG. Synthesis Sci USA 2005; 102: 12129–12134. and characterization of insulin-like growth factor-binding protein (IGFBP)-7. 12 Versteege I, Medjkane S, Rouillard D, Delattre O. A key role of the hSNF5/INI1 Recombinant human mac25 protein specifically binds IGF-I and -II. J Biol Chem tumour suppressor in the control of the G1-S transition of the cell cycle. Oncogene 1996; 271: 30322–30325. 2002; 21: 6403–6412. 39 Wajapeyee N, Serra RW, Zhu X, Mahalingam M, Green MR. Oncogenic BRAF 13 Isakoff MS, Sansam CG, Tamayo P, Subramanian A, Evans JA, Fillmore CM et al. induces senescence and apoptosis through pathways mediated by the secreted Inactivation of the Snf5 tumor suppressor stimulates cell cycle progression and protein IGFBP7. Cell 2008; 132: 363–374. cooperates with p53 loss in oncogenic transformation. Proc Natl Acad Sci USA 40 Figueroa JA, Sharma J, Jackson JG, McDermott MJ, Hilsenbeck SG, Yee D. 2005; 102: 17745–17750. Recombinant insulin-like growth factor binding protein-1 inhibits IGF-I, serum, 14 Cheng SW, Davies KP, Yung E, Beltran RJ, Yu J, Kalpana GV. c-MYC interacts with and estrogen-dependent growth of MCF-7 human breast cancer cells. JCell INI1/hSNF5 and requires the SWI/SNF complex for transactivation function. Nat Physiol 1993; 157: 229–236. Genet 1999; 22: 102–105. 41 Evdokimova V, Tognon CE, Benatar T, Yang W, Krutikov K, Pollak M et al. IGFBP7 15 Jagani Z, Mora-Blanco EL, Sansam CG, McKenna ES, Wilson B, Chen D et al. Loss of binds to the IGF-1 receptor and blocks its activation by insulin-like growth factors. the tumor suppressor Snf5 leads to aberrant activation of the Hedgehog-Gli Sci Signal 2012; 5: ra92. pathway. Nat Med 2010; 16: 1429–1433. 42 Hooper AT, Shmelkov SV, Gupta S, Milde T, Bambino K, Gillen K et al. 16 Klochendler-Yeivin A, Picarsky E, Yaniv M. Increased DNA damage sensitivity and Angiomodulin is a specific marker of vasculature and regulates vascular apoptosis in cells lacking the Snf5/Ini1 subunit of the SWI/SNF chromatin remo- endothelial growth factor-A-dependent neoangiogenesis. Circ Res 2009; 105: deling complex. Mol Cell Biol 2006; 26: 2661–2674. 201–208. 17 Smith MJ, Walker JA, Shen Y, Stemmer-Rachamimov A, Gusella JF, Plotkin SR. 43 Pen A, Durocher Y, Slinn J, M Rukhlova, Charlebois C, DB Stanimirovic et al. Expression of SMARCB1 (INI1) mutations in familial schwannomatosis. Hum Mol Insulin-like growth factor binding protein 7 exhibits tumor suppressive and vessel Genet 2012; 21: 5239–5245. stabilization properties in U87MG and T98G glioblastoma cell lines. Cancer Biol 18 Ae K, Kobayashi N, Sakuma R, Ogata T, Kuroda H, Kawaguchi N et al. Chromatin Ther 2011; 12: 634–646. remodeling factor encoded by ini1 induces G1 arrest and apoptosis in ini1-defi- 44 Chen D, Yoo BK, Santhekadur PK, Gredler R, Bhutia SK, Das SK et al. Insulin-like cient cells. Oncogene 2002; 21: 3112–3120. growth factor-binding protein-7 functions as a potential tumor suppressor in 19 Reincke BS, Rosson GB, Oswald BW, Wright CF. INI1 expression induces cell cycle hepatocellular carcinoma. Clin Cancer Res 2011; 17: 6693–6701. arrest and markers of senescence in malignant rhabdoid tumor cells. J Cell Physiol 45 Tamura K, Hashimoto K, Suzuki K, Yoshie M, Kutsukake M, Sakurai T. Insulin-like 2003; 194: 303–313. growth factor binding protein-7 (IGFBP7) blocks vascular endothelial cell growth 20 Betz BL, Strobeck MW, Reisman DN, Knudsen ES, Weissman BE. Re-expression of factor (VEGF)-induced angiogenesis in human vascular endothelial cells. Eur J hSNF5/INI1/BAF47 in pediatric tumor cells leads to G1 arrest associated with Pharmacol 2009; 610: 61–67. induction of p16ink4a and activation of RB. Oncogene 2002; 21: 5193–5203. 46 Drobic B, Perez-Cadahia B, Yu J, Kung SK, Davie JR. Promoter chromatin remodeling 21 Charboneau A, Chai J, Jordan J, Funkhouser W, Judkins A, Biegel J et al. P-Akt of immediate-early genes is mediated through H3 phosphorylation at either expression distinguishes two types of malignant rhabdoid tumors. J Cell Physiol serine 28 or 10 by the MSK1 multi-protein complex. Nucleic Acids Res 2010; 38: 2006; 209: 422–427. 3196–3208. 22 Jozwiak J, Bikowska B, Grajkowska W, Sontowska I, Roszkowski M, Galus R. 47 Zhang M, Fang H, Zhou J, Herring BP. A novel role of Brg1 in the regulation of Activation of Akt/mTOR pathway in a patient with atypical teratoid/rhabdoid SRF/MRTFA-dependent smooth muscle-specific gene expression. J Biol Chem tumor. Folia Neuropathol 2010; 48: 185–189. 2007; 282: 25708–25716. 23 Foster K, Wang Y, Zhou D, Wright C. Dependence on PI3K/Akt signaling for 48 Xi Q, He W, Zhang XH, Le HV, Massague J. Genome-wide impact of the BRG1 SWI/ malignant rhabdoid tumor cell survival. Cancer Chemother Pharmacol 2009; 63: SNF chromatin remodeler on the transforming growth factor beta transcriptional 783–791. program. J Biol Chem 2008; 283: 1146–1155. 24 Nicholson KM, Anderson NG. The protein kinase B/Akt signalling pathway in 49 Nocentini S. Apoptotic response of malignant rhabdoid tumor cells. Cancer Cell Int human malignancy. Cell Signal 2002; 14: 381–395. 2003; 3:11. 25 Datta SR, Brunet A, Greenberg ME. Cellular survival: a play in three Akts. Genes Dev 50 Arcaro A, Doepfner KT, Boller D, Guerreiro AS, Shalaby T, Jackson SP et al. Novel 1999; 13: 2905–2927. role for insulin as an autocrine growth factor for malignant brain tumour cells. 26 Brunet A, Bonni A, Zigmond MJ, Lin MZ, Juo P, Hu LS et al. Akt promotes cell Biochem J 2007; 406: 57–66. survival by phosphorylating and inhibiting a Forkhead transcription factor. Cell 51 Vizioli MG, Sensi M, Miranda C, Cleris L, Formelli F, Anania MC et al. IGFBP7: 1999; 96: 857–868. an oncosuppressor gene in thyroid carcinogenesis. Oncogene 2010; 29: 27 Burgering BM, Coffer PJ. Protein kinase B (c-Akt) in phosphatidylinositol-3-OH 3835–3844. kinase signal transduction. Nature 1995; 376: 599–602. 52 Sato Y, Chen Z, Miyazaki K. Strong suppression of tumor growth by insulin-like 28 Ruvinsky I, Sharon N, Lerer T, Cohen H, Stolovich-Rain M, Nir T et al. Ribosomal growth factor-binding protein-related protein 1/tumor-derived cell adhesion protein S6 phosphorylation is a determinant of cell size and glucose homeostasis. factor/mac25. Cancer Sci 2007; 98: 1055–1063. Genes Dev 2005; 19: 2199–2211. 53 Benatar T, Yang W, Amemiya Y, Evdokimova V, Kahn H, Holloway C et al. IGFBP7 29 Blume-Jensen P, Hunter T. Oncogenic kinase signalling. Nature 2001; 411: 355–365. reduces breast tumor growth by induction of senescence and apoptosis path- 30 Lee RS, Stewart C, Carter SL, Ambrogio L, Cibulskis K, Sougnez C et al. ways. Breast Cancer Res Treat 2012; 133: 563–573. A remarkably simple genome underlies highly malignant pediatric rhabdoid 54 Burger AM, Leyland-Jones B, Banerjee K, Spyropoulos DD, Seth AK. Essential roles cancers. J Clin Invest 2012; 122: 2983–2988. of IGFBP-3 and IGFBP-rP1 in breast cancer. Eur J Cancer 2005; 41: 1515–1527. 31 Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, Gillette MA et al. 55 Yamanaka Y, Wilson EM, Rosenfeld RG, Oh Y. Inhibition of insulin receptor acti- Gene set enrichment analysis: a knowledge-based approach for interpreting vation by insulin-like growth factor binding proteins. J Biol Chem 1997; 272: genome-wide expression profiles. Proc Natl Acad Sci USA 2005; 102: 30729–30734. 15545–15550. 56 Ahmed S, Yamamoto K, Sato Y, Ogawa T, Herrmann A, Higashi S et al. Proteolytic 32 Mootha VK, Lindgren CM, Eriksson KF, Subramanian A, Sihag S, Lehar J et al. processing of IGFBP-related protein-1 (TAF/angiomodulin/mac25) modulates its PGC-1alpha-responsive genes involved in oxidative phosphorylation are biological activity. Biochem Biophys Res Commun 2003; 310: 612–618. coordinately downregulated in human diabetes. Nat Genet 2003; 34: 267–273. 57 Beattie J, Kreiner M, Allan GJ, Flint DJ, Domingues D, van der Walle CF. IGFBP-3and 33 Wang L, Baiocchi RA, Pal S, Mosialos G, Caligiuri M, Sif S. The BRG1- and hBRM- IGFBP-5 associate with the cell binding domain (CBD) of fibronectin. Biochem associated factor BAF57 induces apoptosis by stimulating expression of the Biophys Res Commun 2009; 381: 572–576. cylindromatosis tumor suppressor gene. Mol Cell Biol 2005; 25: 7953–7965. 58 Collett-Solberg PF, Cohen P. The role of the insulin-like growth factor binding 34 Hendricks KB, Shanahan F, Lees E. Role for BRG1 in cell cycle control and tumor proteins and the IGFBP proteases in modulating IGF action. Endocrinol Metab Clin suppression. Mol Cell Biol 2004; 24: 362–376. North Am 1996; 25: 591–614.

& 2014 Macmillan Publishers Limited Oncogene (2014) 3024 – 3032 AKT activation following SMARCB1 loss J Darr et al 3032 59 Hollmann TJ, Hornick JL. INI1-deﬁcient tumors: diagnostic features and molecular 62 Muchardt C, Sardet C, Bourachot B, Onufryk C, Yaniv M. A human protein with genetics. Am J Surg Pathol 2011; 35: e47–e63. homology to Saccharomyces cerevisiae SNF5 interacts with the potential helicase 60 Euhus DM, Hudd C, LaRegina MC, Johnson FE. Tumor measurement in the nude hbrm. Nucleic Acids Res 1995; 23: 1127–1132. mouse. J Surg Oncol 1986; 31: 229–234. 63 Barzily-Rokni M, Friedman N, Ron-Bigger S, Isaac S, Michlin D, Eden A. Synergism 61 Naviaux RK, Costanzi E, Haas M, Verma IM. The pCL vector system: rapid production between DNA methylation and macroH2A1 occupancy in epigenetic silencing of of helper-free, high-titer, recombinant retroviruses. J Virol 1996; 70: 5701–5705. the tumor suppressor gene p16 (CDKN2A). Nucleic Acids Res 2011; 39: 1326–1335.

Supplementary Information accompanies this paper on the Oncogene website (http://www.nature.com/onc)

Oncogene (2014) 3024 – 3032 & 2014 Macmillan Publishers Limited