SATB2 Enhances Migration and Invasion in Osteosarcoma by Regulating Genes Involved in Cytoskeletal Organization

ORIGINAL ARTICLE SATB2 enhances migration and invasion in osteosarcoma by regulating genes involved in cytoskeletal organization

BKA Seong1,2, J Lau2, T Adderley2, L Kee2, D Chaukos2, M Pienkowska3, D Malkin1,3,4, P Thorner5,6 and MS Irwin1,2,4,6

Osteosarcoma (OS) is the most common malignant bone tumor and the majority of recurrences are due to metastasis. However, the molecular mechanisms that regulate OS metastatic spread are largely unknown. In this study, we report that special AT-rich-binding protein 2 (SATB2) is highly expressed in OS cells and tumors. Short hairpin RNA-mediated knockdown of SATB2 (sh-SATB2) decreases migration and invasion of OS cells without affecting proliferation or viability. Microarray analysis identified genes that were differentially regulated by SATB2 including the actin-binding protein Epithelial Protein Lost In Neoplasm (EPLIN), which was upregulated in sh-SATB2 cells. Silencing EPLIN rescues the decreased invasion observed in sh-SATB2 cells. Pathway analyses of SATB2-regulated genes revealed enrichment of those involved in cytoskeleton dynamics, and increased stress fiber formation was detected in cells with SATB2 knockdown. Furthermore, sh-SATB2 cells exhibit increased RhoA, decreased Rac1 and increased phosphorylation of focal adhesion kinase (FAK) and paxillin. These findings identify SATB2 as a novel regulator of OS invasion, in part via effects on EPLIN and the cytoskeleton.

Oncogene (2015) 34, 3582–3592; doi:10.1038/onc.2014.289; published online 15 September 2014

INTRODUCTION cancer cells showed enrichment in genes involved in growth, Osteosarcoma (OS) is the most common malignant bone tumor in including cell cycle and phosphatidylinositol signaling, and children and adolescents and arises from mesenchymal cells of metastasis, such as cell adhesion, extracellular matrix and focal 6 osteoblast (OB) origin. Despite advances in surgery and che- adhesions. Our group previously reported the first role for SATB2 motherapy, survival rates remain 65–70%.1 For patients with in cancer, identifying SATB2 as a novel binding partner for p63 metastases, most commonly pulmonary, survival is o20%. and p73, but not for p53. In head and neck squamous cell Although germline mutations in Rb and p53 are associated with carcinoma (HNSCC), SATB2 augments ΔNp63α activity to promote the development of OS, the pathways that regulate OS metastasis chemoresistance and SATB2 expression was associated with more 7 are not well understood. Many proteins with roles in bone advanced stage HNSCC. SATB2 has also been reported to be development have been implicated in OS. Mice with mutant upregulated in cancer-associated fibroblasts mediating migration Special AT-rich-binding protein 2 (SATB2) have significant defects of endometrial cancer cells,8 and higher SATB2 mRNA levels were in bone development and differentiation.2,3 In this study, we detected in breast cancer tissue compared with normal matched identify a role for SATB2 as a regulator of OS cell migration and tissue.9 In contrast, SATB2 expression is associated with more invasion. favorable outcome in colon carcinoma,10 suggesting that like SATB2 is a member of the SATB family of transcription factors, many other proteins involved in tumorigenesis, the function(s) of which share structural homology consisting of PDZ, CUT and SATB2 in cancer may be cell or context dependent and further homeobox domains. They bind to AT-rich DNA sequences in studies are required to dissect the molecular mechanisms by nuclear matrix attachment regions to regulate gene expression by which SATB2 may promote tumorigenesis and/or invasion and orchestrating chromatin organization and remodeling.4 In addi- metastasis. tion, SATB2 can directly bind transcription factors acting as a Regulation of the cytoskeleton is critical during metastasis, a co-activator or a repressor. Knockout mouse models have multi-step process involving cancer cell migration, invasion, demonstrated roles for SATB2 in craniofacial morphogenesis and survival, adhesion and colonization. The actin cytoskeleton forms OB differentiation in part via SATB2-mediated repression of Hoxa2 filopodia and lamellapodia at the leading edge of migrating cells and cooperation with ATF4 and Runx2 to promote OB and mediates extracellular signals into the cell via integrins and differentiation.3 SATB2 also regulates neuronal specification and other cell surface receptors.11 The dynamics of the actin migration by recruiting chromatin-remodeling complexes to Ctip2 cytoskeleton are regulated by small GTPases including RhoA and locus.5 However, the role of SATB2 in cancer is less well Rac1, and actin-binding proteins such as Epithelial Protein Lost In understood. Interestingly, the SATB2 homolog SATB1 promotes Neoplasm (EPLIN) and cofilin. For example, the activation of Rac1 breast tumor growth and metastasis by reprogramming global promotes membrane ruffling and lamellapodia formation, gene expression.6 Pathway analyses of SATB1-depleted breast whereas RhoA regulates stress fiber formation and adhesion via

1Departments of Pediatrics and Medical Biophysics, University of Toronto, Toronto, Ontario, Canada; 2Cell Biology Program, Hospital for Sick Children Research Institute, Peter Gilgan Centre for Research and Learning, Toronto, Ontario, Canada; 3Program in Genetics and Genome Biology, Hospital for Sick Children Research Institute, Peter Gilgan Centre for Research and Learning, Toronto, Ontario, Canada; 4Division of Hematology-Oncology, Department of Pediatrics, Hospital for Sick Children, Toronto, Ontario, Canada; 5Department of Pediatric Laboratory Medicine, Hospital for Sick Children, Toronto, Ontario, Canada and 6Department of Laboratory Medicine and Pathobiology, University of Toronto, Toronto, Ontario, Canada. Correspondence: Dr M Irwin, Department of Paediatrics and Cell Biology, The Hospital for Sick Children, 555 University Avenue, Black Wing, 9th floor, Toronto, Ontario, Canada M5G 1X8. E-mail: [email protected] Received 4 February 2014; revised 28 July 2014; accepted 31 July 2014; published online 15 September 2014 The role of SATB2 in osteosarcoma invasion BKA Seong et al 3583 the activity of downstream kinases (for example, Rho-associated expression in 41/44 (93%) OS samples. In contrast, SATB2 kinase, ROCK).12 EPLIN, a cytoskeleton-associated protein that expression was undetectable in 76/77 of non-OS sarcomas binds to and stabilizes actin filaments, is downregulated in many (Figure 1c). cancers such as prostate and esophageal, and has been shown to negatively regulate growth and invasion by inhibiting epithelial- SATB2 knockdown decreases migration and invasion of OS cells to-mesenchymal transition.13 In epithelial cells, EPLIN links cadherin − catenin complexes to F-actin, stabilizing apical adhe- To determine the role of SATB2 in OS, we generated OS cell lines sion belts and also inhibits Rac1-induced ruffling by cross-linking with stable knockdown of SATB2 (KHOS-shSATB2-1 and KHOS- and forming stabilized F-actin structures such as stress fibers at shSATB2-2) and control (KHOS-shGFP). shSATB2-1 and -2 target 7 fi the expense of more dynamic actin filament structures.14,15 The different SATB2 sequences. Knockdown was con rmed by role of EPLIN and how it is regulated in OS has not been previously immunoblot (Figure 2a). Since the level of SATB2 was higher in described. Here, we demonstrate that SATB2 is highly expressed in murine pulmonary metastases in comparison with matched and is a specific marker for OS, and that SATB2 regulates the primary OS tumors we predicted that SATB2 might be involved expression of genes involved in actin cytoskeleton dynamics to in metastastic spread via effects on migration and/or invasion. promote OS invasion and migration, including EPLIN, which is a Significant decreases in migration and invasion were observed in key mediator by which SATB2 regulates OS invasion. three KHOS-shSATB2 cell lines compared with control cells, as detected using Boyden chambers (Figure 2a; Supplementary Figure 5a). To exclude the possibility that decreased invasion in RESULTS shSATB2 knockdown cells may reflect differences in proliferation, SATB2 is highly expressed in OS we performed BrdU and MTT assays. We did not detect any We had previously reported that high levels of SATB2 are detected differences in proliferation or viability between KHOS-shSATB2 in advanced stage HNSCC primary tumors and cell lines.7 We cells, -sh-control and parental control cells, suggesting that the screened more than 20 cancer cell lines from different tumors and effects observed in response to SATB2 knockdown were due to noted high SATB2 protein levels in OS cell lines (data not shown). migration and invasion (Figure 2b). Similarly, we also performed SATB2, but not SATB1, was detected in all five human OS cell lines scratch-wound migration assay to look at collective cell migration tested at levels significantly higher than those detected in non- and detected decreased migration in KHOS-shSATB2 cells transformed hOBs (human OB s), the putative cell of origin for OS compared with control (Figure 2c). To exclude cell line- or clone- (Figure 1a). The increased expression in OS cell lines in comparison specific effects, we performed transient SATB2 knockdown in with non-transformed OBs suggests that SATB2 may be involved MNNG and U2OS cells and observed that lower levels of SATB2 in transformation of OBs to OS cells. We also examined SATB2 were associated with decreased migration and invasion (Figure 2a expression in murine OS cells that were derived from the MOTO and data not shown). Since knockdown led to decreased invasion mouse model of OS in which expression of the simian virus we examined whether SATB2 overexpression would result in T-antigen transgene is driven by an OB-specific promoter.16 SATB2 increased invasion. KHOS cells were transfected with a plasmid was expressed in both primary bone tumors and lung metastases. encoding T7-tagged SATB2 to generate two cell lines with stable Interestingly, SATB2 expression was higher in cells derived from SATB2 overexpression: KHOS-T7-SATB2-1 and KHOS-T7-SATB2-2. metastases compared with matched primary tumors (Figure 1b), As predicted, overproduction of SATB2 led to increased migration suggesting that SATB2 may have a role in promoting OS and invasion compared with the control cells (Figure 2d). metastasis in vivo. We next asked whether SATB2 is expressed in Furthermore, the relative level of migration and invasion observed primary human OS tumors. Immunostaining of a tissue microarray in OS cell lines correlated with endogenous SATB2 expression of primary pediatric tumors showed nuclear SATB2 protein (Figure 1a; Supplementary Figure 1).

Matched

U2OS

hOB

HOS

SAOS-2

MNNG

Ad-SATB1 Ad-SATB2 HaCaT

KHOS Primary SATB2 Met Primary Met SATB2 SATB1

β-actin Vinculin

Type of Sarcoma SATB2+ SATB2-

Control SATB2 Osteosarcoma 41 3 Rhabdomyosarcoma 0 32 Ewing’s sarcoma 0 33 Synovial sarcoma 0 6 Neurofibrosarcoma 1 6

Figure 1. SATB2 expression in OS. (a) Lysates from human OB and OS cell lines were immunoblotted for SATB1 and SATB2. HaCat cells transduced with adenovirus encoding SATB1 or SATB2 were used as positive controls. (b) Lysates of Murine OS cell lines derived from primary tumors and pulmonary metastases of MOTO mouse were immunoblotted for SATB2. Note: hOB lysate was resolved in the same gel as murine OS cells. (c) OS tumor biopsy samples were immunostained with SATB2 antibody. Examples of positive (brown nuclear staining) and negative tumors are shown. Chart: Results of SATB2 immunostaining of tissue microarray with different histologic types of non-OS sarcomas.

a MIG MIG INV 1.4 1.4 INV

1.2 1.2

1.0 1

0.8 0.8

MNNG- shSATB2-2 MNNG- shGFP MNNG- shSATB2-1

shSATB2-2 shSATB2-1 shGFP

KHOS KHOS- KHOS- KHOS- 0.6 0.6

SATB2 Change Fold 0.4 SATB2 0.4

0.2 0.2 Vinculin Vinculin 0 0 KHOS- KHOS- KHOS- MNNG- MNNG- MNNG- shGFP shSATB2-1 shSATB2-2 shGFP shSATB2-1 shSATB2-2

b MTT BrdU 2.5 3.5 KHOS KHOS KHOS- shGFP 3 KHOS- shGFP 2 KHOS- shSATB2-1 KHOS- shSATB2-1 KHOS- shSATB2-2 2.5 KHOS- shSATB2-2

1.5 2

1.5

1 Abs @ 450 nm 450 @ Abs Abs @ 575 nm 575 @ Abs 1 0.5 0.5

0 0 24 48 72 24 48 72 Time (h) Time (h)

c 0 h 6 h 12 h 24 h

KHOS- shGFP

KHOS- shSATB2-1

KHOS- shSATB2-2

d MIG 1.8 INV 1.6 1.4 1.2 1

KHOS KHOS- T7-SATB2-1 KHOS- T7-SATB2-2 0.8

T-7 tag Change Fold 0.6 0.4 Vinculin 0.2 0 KHOS KHOS- KHOS- T7-SATB2-1 T7-SATB2-2 Figure 2. SATB2 knockdown decreases OS invasion and migration. (a) SATB2 immunoblot of lysates from KHOS cell lines with stable knockdown with the indicated lenti-shRNAs (shGFP control, shSATB2-1 (bp 506), shSATB2-2 (bp 1335)). Transwell migration (MIG) assay using Boyden chambers. shSATB2 and control KHOS cells (left panel), and MNNG cells transiently infected with indicated shRNAs (right panel) were assessed for migration using FBS gradient 20 h after seeding. Matrigel was used for invasion (INV) assays. Cells were lysed and immunoblotted with anti-SATB2 for SATB knockdown. The data represent fold change normalized to the control ± s.d. (N = 3; *Po0.05) (b) MTT and BrdU assays were performed for sh-SATB2 and controls cells (in triplicate) for the indicated times. Data represent mean ± s.d. (N = 3). (c) Scratch-wound assay was performed on conﬂuent monolayers of sh-SATB2 and control cells. Representative images at indicated time points are shown. The data represent mean. Graph demonstrates % recovery of wounds in cells with sh-SATB2 (N = 3). (d) T-7 immunoblot of lysates of KHOS cells that were transfected and selected for SATB2 overexpression (transfected with plasmid encoding T7-SATB2). KHOS-T7-SATB2 and parental cells were subjected to transwell migration and invasion assays described above. Data represent fold change normalized to the control ± s.d. (N = 3; *Po0.05).

Upregulated Genes 15 ADAM23 CDH6 E2F5 EFEMP1 EGFRGPR126 KRT8LIMA1 MMP16 NNMT ESAM ENPP1 NOX4 ROBO4

KHOS- KHOS- 10 shSATB2-1 shSATB2-2 CXCL10 6.733 MMP16 9.697 KRT8 6.366 GRP126 4.645 5 FAM38B 6.000 SLAIN1 4.183 EFEMP1 5.958 EFEMP1 3.815 0

MMP16 5.853 HBE1 3.572 (ddCT) CDH6 5.401 CDH3 3.519 NNMT 5.229 SDC2 3.504 -5 PRF126 4.622 MYPN 3.409 HBE1 4.092 FAM38B 3.149 Change -10 KHOS- shSATB2-1 LIMA1 3.881 E2F5 3.125 KHOS- shSATB2-2 Downregulated Genes -15 Array shSATB2-1

Avg.Fold Avg.Fold Array shSATB2-2 KHOS- KHOS- shSATB2-1 shSATB2-2 -20 MUM1L1 -4.226 TM4SF18 -10.43 ENPP1 -4.193 SPP1 -6.873 -25 NOX4 -3.588 MUM1L1 -6.580 ITGA8 -3.400 ZNF626 -5.972 MFAP2 -2.690 FABP4 -5.112 PLXNA2 -2.687 IFI44 -4.733 ESAM -2.665 CD96 -4.576 PRRX1 -2.661 NOX4 -4.415 PDE3A -2.641 PDE3A -3.684

ROBO4 -2.605 FLVCR2 -3.663

ATB2-2

shGFP shGFP shSATB2-1 shSATB2-1 shSATB2-2 shS

KHOS- KHOS- KHOS- KHOS- KHOS- KHOS-

ENPP1 MMP16

SATB2 SATB2

α-Tubulin Vinculin

ATB2-2

shGFP shSATB2-1 shSATB2-2 shGFP shSATB2-1

KHOS- KHOS- KHOS- KHOS- KHOS-

shS KHOS-

KRT8 NOX4

SATB2 SATB2

Vinculin Vinculin

Enriched Pathways Enrichment (KEGG-based) P-value Regulation of Actin Cytoskeleton 2.92E-04 Focal Adhesion 1.10E-03 ECM-Receptor Interaction 0.001133 PI3K-Akt Signaling Pathway 0.017219 Cell Adhesion Molecules (CAMs) 0.031892

Figure 3. SATB2 changes global gene expression. (a) Ten genes with the most significant differential expression in sh-SATB2-1 and -2 cells compared with sh-Control. (b) Validation of the microarray by real-time PCR. Values normalized to shGFP control. The data represent average fold change (ddCT) ± s.d. (N = 3). (c) Validation of the microarray by immunoblot analysis. shSATB2 and shControl cells were lysed and immunoblotted for proteins that showed differential mRNA expression in shSATB2 cells compared with shControl. (d) Cytoscape view of the top enriched pathways in shSATB2 cells compared with shControl using GSEA (Po0.05; Qo0.1). (e) List of top five enriched pathways in shSATB2 cells compared with the control revealed by KEGG-based pathway enrichment analysis performed with the list of genes filtered for fold change ± 1.5 and Po0.05.

Identiﬁcation of genes regulated by SATB2 shRNA, shSATB2-1 and shSATB2-2 cells using the Affymetrix Given the role of SATB family proteins in chromatin remodeling Human Gene platform. Approximately 425 genes were signiﬁ- and as transcription factors, we asked whether SATB2 might cantly differentially regulated between KHOS-shSATB2 and control regulate the expression of genes involved in migration and cells. The most differentially expressed genes based on fold invasion. We performed a microarray analysis with KHOS control changes in two different clones of sh-SATB2 cells in comparison

© 2015 Macmillan Publishers Limited Oncogene (2015) 3582 – 3592 The role of SATB2 in osteosarcoma invasion BKA Seong et al 3586 with control cells include LIMA1, MMP16, GPR126, EFEMP1, partially rescue the decreased invasion observed in sh-SATB2 cells. MUM1L1 and NOX4 (Figure 3a). To validate our microarray data, Thus, we asked whether the expression of SATB2 was associated we performed qRT-PCR and immunoblot analyses to confirm with global changes in OS cell cytoskeleton. Immunofluorescence expression of candidate SATB2-target genes that we prioritized with phalloidin staining of actin filaments revealed increased based on (1) fold difference between control and SATB2 knock- stress fibers in sh-SATB2 cells compared with the control cells down cells, (2) similar results in both knockdown lines, (Figure 5a; Supplementary Figures 2 and 5b). In contrast, we did (3) potential role in OS or metastases (Figures 3b and c). not observe any differences in microtubule structures as detected To determine whether differentially expressed genes had by tubulin staining (Supplementary Figure 3). The actin cytoske- functional relationships in similar signaling pathways, we leton is highly regulated by the activities of small GTPases performed pathway-based analyses using GSEA (Gene Set including RhoA, Rac1 and Cdc42, which act as molecular switches Enrichment Analysis) tool and Partek software (KEGG based). controlling multiple signaling pathways. These proteins are GSEA revealed enrichment of genes in pathways such as activated by GTP binding and are inactivated by the hydrolysis cytoskeleton organization, cell leading edge and regulation of of GTP to GDP. Since it is well known that RhoA activity regulates small GTPase signaling, all of which are associated with cell stress fiber formation and adhesion,12,20 we asked whether sh- migration (Figure 3d). In agreement with roles in chromatin SATB2 cells with increased stress fibers have altered RhoA remodeling, the SATB differentially expressed genes were also activation. GTPase activation assay using rhotekin-RBD beads to enriched for pathways involved in chromosome and chromatin pull-down GTP-bound RhoA demonstrated increased RhoA organization and modification (Figure 3d). The KEGG-based activation in sh-SATB2 cells compared with the control pathway enrichment analysis of genes that were differentially (Figure 5b; Supplementary Figure 5f). Under the same conditions expressed by 41.5-fold revealed that the greatest proportion of we observed decreased Rac1 activation in sh-SATB2 cells genes were associated with regulation of actin cytoskeleton (Figure 5b). To further determine which effector of RhoA may followed by focal adhesion, and extracellular matrix–receptor mediate the migration phenotype, we treated sh-SATB2 cells with interaction (Figure 3e), all of which are also known to regulate a pan-ROCK (Y-27632) and an ROCK2-specific inhibitor. Pharma- cytoskeleton dynamics required for cell migration. For example, cologic inhibition of ROCK1, but not ROCK2, rescued the ABI3, ABI3BP, GSN and WASH1 genes, which promote actin decreased migration observed in sh-SATB2 cells without affecting polymerization and branching that is crucial for cell proliferation (Supplementary Figure 4). This suggests that ROCK1 migration,17–19 were downregulated in KHOS-shSATB2 cells. is an effector of RhoA that regulates migration in OS cells. We next examined cell adhesion by staining adherent cells at The SATB2 regulated gene EPLIN mediates the decreased different time points and observed decreased adhesion in sh- migration and invasion in SATB2 knockdown cells SATB2 cells compared with the control (Figure 5c; Supplementary Figure 5c). Consistent with the role of RhoA in driving maturation Lima1 was one of the most highly upregulated genes in SATB2- of focal adhesions, we observed increased phosphorylation of knockdown cells. Lima1 encodes EPLIN, which has two actin- focal adhesion kinase (FAK) and paxillin in sh-SATB2 cells upon binding domains flanking a central LIM domain. Microarray serum induction (Figure 5d; Supplementary Figure 5d). As analysis showing upregulated EPLIN expression in sh-SATB2 was confirmation, immunofluorescence showed increased focal adhe- validated at both the RNA and protein levels using real-time PCR sions in sh-SATB2 cells compared with the control as measured by (RT-PCR), immunoblot and immunofluorescence assays (Figures p-FAK and p-paxillin staining (Figure 5e; Supplementary Figure 3b and 4a). To exclude cell line- and shRNA-specific effects, we 5e). It has previously been shown that EPLIN inhibits Rac-1 transiently transfected U2OS cells with siRNA targeting SATB2 and induced ruffling and increases stress fibers by cross-linking and observed increased EPLIN expression (Figure 4b). This inverse bundling actin filaments,21 however, the functional link between relationship between SATB2 and EPLIN expression was also EPLIN and adhesion is poorly understood. Thus, we examined observed at the endogenous level (Figure 4b). Lysates from cells whether EPLIN also regulates SATB2-dependent adhesion in OS with high SATB2 levels had lower levels of EPLIN. To determine cells. EPLIN knockdown in sh-SATB2 cells resulted in increased whether SATB2 directly regulates EPLIN expression, we examined adhesion, partially rescuing the decreased adhesion detected in whether the EPLIN promoter contains consensus SATB-binding sh-SATB2 cells (Figure 6a). In addition, EPLIN knockdown resulted site. BLAST analysis revealed no SATB-binding sites in promoters in decreased stress fiber formation in the sh-SATB2 cells of both α and β isoforms of EPLIN, suggesting that the regulation (Figure 6b). To further elucidate the mechanism by which EPLIN of EPLIN expression by SATB2 is not due to direct SATB2 binding regulates adhesion, we examined focal adhesion proteins. to the EPLIN promoter (data not shown). Transient knockdown of EPLIN expression in sh-SATB2 cells The role of EPLIN in regulating actin cytoskeleton, together with decreased paxillin expression and serum-induced phosphorylation our data supporting the role of SATB2 in OS cell invasion, led us to of paxillin (Figure 6c). In addition, EPLIN overexpression was investigate the role of EPLIN in SATB2-mediated invasion. Given associated with increased phosphorylation of paxillin and that EPLIN is upregulated in sh-SATB2 cells, we asked whether decreased invasion (Figures 6d and e). This finding suggests that inhibiting EPLIN expression in sh-SATB2 cells, which already have EPLIN regulates both the phosphorylation and turnover of paxillin increased EPLIN expression, can rescue the decreased invasion and invasion. phenotype. To this end, sh-SATB2 cells transfected with siRNA Taken together, our data support a model whereby high levels targeting EPLIN were subjected to migration and invasion assays. of SATB2 in OS lead to EPLIN downregulation, which in turn results Downregulating EPLIN expression resulted in increased migration in decreased phosphorylated paxillin, increased adhesion and and invasion of sh-SATB2 cells compared with the control cells, decreased stress fibers, resulting in increased OS cell invasion partially rescuing the sh-SATB2 decreased invasion phenotype (Figure 6f). (Figure 4c), suggesting that SATB2-mediated invasion is at least in part due to effects on EPLIN expression. DISCUSSION SATB2 modulates actin cytoskeleton and effects are mediated by The SATB family proteins have diverse functions due to their EPLIN abilities to regulate gene expression and higher order chromatin Our pathway analyses revealed enrichment of genes that regulate structure. Until recently, the majority of reports on SATB1 and the actin cytoskeleton, which is essential for cell migration, and SATB2 were limited to roles in development of the immune and knockdown of EPLIN, an actin binding protein, was able to nervous systems, respectively,22,23 and roles for SATB proteins in

KHOS- KHOS- KHOS KHOS shSATB2-1 KHOS- shSATB2-2 KHOS- T7-SATB2-1 KHOS- T7-SATB2-2 KHOS- shGFP shGFP shSATB2-1 SATB2

EPLIN EPLIN

Vinculin

U2OS

MNNG HOS

siSATB2 (nM) SAOS-2 KHOS siControl U2OS 7.5 75 SATB2

SATB2 EPLIN (Long exposure)

EPLIN EPLIN (Short exposure)

Vinculin Vinculin

* MIG 1.2 INV shGFP + - - - 1 shSATB2-2 - + + + siControl - - + - 0.8 siEPLIN - - - + 0.6

EPLIN 0.4 Fold Change Fold 0.2 SATB2 0 + - - - shGFP - + + + shSATB2-2 Vinculin - - + - siControl - - - + siEPLIN Figure 4. EPLIN is downregulated by SATB2 and EPLIN knockdown partially rescues shSATB2 MIG/INV phenotype. (a) Lysates of KHOS- shSATB2, KHOS-T7-SATB2 clones and control cells immunoblotted with antibodies to SATB2 and EPLIN (left panel). KHOS-shSATB2-2 and control cells were plated on ﬁbronectin-coated slides, ﬁxed and stained with EPLIN antibody. Representative images of N = 3 experiments (right panel) are shown. (b) Lysates of U2OS cells transiently transfected (48 h) with indicated siRNAs (left panel) and lysates of various OS cell lines (right panel) immunoblotted with antibodies to SATB2 and EPLIN. (c) KHOS-shSATB2-2 stable cells were transiently transfected with control siRNA or siRNA targeting EPLIN and subjected to transwell migration and invasion assay. Data represent fold change normalized to KHOS-shGFP control ± s.d. (N = 3; *Po0.05).

cancer had not been well described. Han et al.6 demonstrated that Previous reports of patients and mice with mutant SATB2 SATB1 is highly expressed in aggressive breast tumors and demonstrate a critical role in OB differentiation.2,3,24,25 SATB2 is reprograms gene expression to promote growth and metastases. required for early OB differentiation and its expression is turned We recently showed that SATB2 is highly expressed in advanced off during terminal stages of differentiation by miR-34b/c and HNSCC where it promotes survival and chemoresistance.7 We now 23a ~ 27a ~ 24-2 clusters.25,26 Here, we demonstrate that the have identiﬁed a novel role for SATB2 in promoting OS invasion expression of SATB2 is higher in OS cells and tumors compared and migration, but not proliferation, by regulating the expression with normal OBs, the putative cell of origin for OS. This suggests of EPLIN and other genes involved in motility, cytoskeletal that OB progenitors expressing high levels of SATB2 either may organization and adhesion. have failed to undergo normal differentiation or may have

KHOS- shGFP

KHOS- shSATB2-2

Adhesion 0.7 KHOS- 0.6 shGFP KHOS- 0.5 shSATB2-2 KHOS- shGFP KHOS- shSATB2-2 KHOS- shGFP KHOS- shSATB2-2 0.4

RhoA-GTP Rac1-GTP 0.3

Abs @ 570 nm 570 @ Abs 0.2

Total RhoA Total Rac1 0.1

0 10 20 30 Time (min)

KHOS- KHOS- shGFP shSATB2-2 DAPI p-Paxillin p-FAK Merge Enlarged p-FAK

KHOS- FAK shGFP

p-Paxillin

KHOS- Paxillin shSATB2-2

Tubulin

Figure 5. SATB2 knockdown affects cytoskeleton. (a) KHOS-shSATB2-2 cells display more stress fibers and stable actin structures compared with the control. KHOS-shSATB2-2 and control cells were plated on fibronectin-coated slides, fixed and stained with AlexaFluor-568 Phalloidin antibody. Representative images of N = 3 experiments are shown. (b) GTPase activation assay demonstrates alterations in RhoA/Rac1 in shSATB2 cells. Serum starved KHOS-shSATB2 and control cells induced with serum were lysed and Rhotekin and PAK beads were used to pull down RhoA-GTP and Rac1-GTP, respectively. Lysates were immunoblotted with anti-RhoA and anti-Rac1. N = 3. (c) Diminished adhesion in shSATB2 cells. KHOS-shSATB2-2 and control cells were seeded, washed, fixed at indicated time points, and stained with crystal violet dye. Data represent absorbance at 570 nm ± s.d. (N = 3; *Po0.05). (d) Increased phosphorylation of FAK and paxillin in shSATB2 cells. KHOS-shSATB2-2 and control cells were serum-starved overnight, induced with serum, lysed and immunoblotted with indicated antibodies. N = 3. (e) Cells lacking SATB2 display more focal adhesions compared with control cells. KHOS-shSATB2-2 and sh-control cells were plated on fibronectin- coated slides, fixed and stained with p-FAK and p-Paxillin antibodies. Representative images of N = 3 experiments are shown. KHOS-shSATB2-2 cells display more focal adhesions compared with the control.

undergone additional genetic events leading to malignant different SATB2 antibody, concluded that SATB2 positivity can transformation, both of which may partially be due to high SATB2 distinguish tumors containing cartilage from those with osteoid. expression. We also showed that SATB2 is a highly speciﬁc marker OS histologic diagnosis is based on morphology and the presence for OS (93% of OS tumors vs 1% of non-OS pediatric sarcomas). of malignant osteoid. The addition of SATB2 immunostaining may The use of SATB2 as an immunohistochemical maker for OS has facilitate diagnosis for OS tumors with minimal osteoid formation also recently been reported by Conner and Homick27 who, using a or other atypical features. Although the molecular mechanism by

Adhesion siEPLIN-2 siEPLIN-1 1.4 siControl siControl 1.2 EPLIN 1 0.8 p-Paxillin 0.6

0.4 Paxillin Abs @ 570 nm 570 @ Abs 0.2 siEPLIN 0 β-actin siControl siEPLIN-1 siEPLIN-2

1.2 MIG INV EPLIN 1 pcDNA3

0.8 pcDNA3 EPLIN EPLIN 0.6 EPLIN

0.4

p-Paxillin Fold Change 0.2 α-Tubulin

Paxillin 0 pcDNA3 EPLIN

p-Paxillin EPLIN

Np73/63? Adhesion

SATB2 Invasion

GEFs GAPs Rho/ROCK Stress Fibres Other SATB2 Targets? Rac1 Ruffling

Figure 6. EPLIN knockdown rescues shSATB2 adhesion phenotype and regulates paxillin. (a) Increased adhesion in sh-SATB2 cells with EPLIN knockdown. KHOS-shSATB2 cells in which EPLIN expression was inhibited with EPLIN siRNA were seeded, washed, fixed at indicated time points, and stained with crystal violet dye. Data represent absorbance at 570 nm ± s.d. (N = 3; *Po0.05). (b) Decreased stress fibers in sh-SATB2 cells with EPLIN knockdown. KHOS-shSATB2 cells in which EPLIN expression was inhibited with EPLIN siRNA were seeded, washed and fixed on fibronectin-coated slides, and stained with phalloidin. Representative images of N = 2 experiments are shown. (c) Decreased stability and phosphorylation of paxillin in shSATB2 cells EPLIN knockdown. KHOS-shSATB2 cells in which EPLIN expression was transiently inhibited with EPLIN siRNA were serum-starved overnight, induced with serum, lysed and immunoblotted with indicated antibodies. N = 3. (d) EPLIN overexpression increases phosphorylation of paxillin. KHOS-shGFP control cells were transiently transfected with increasing amounts of a plasmid encoding EPLIN, lysed and immunoblotted with indicated antibodies. N = 3. (e) EPLIN overexpression increases migration/invasion of KHOS-shGFP control cells. KHOS-shGFP control cells transiently transfected with a plasmid encoding EPLIN were subjected to Boyden chamber migration/invasion assays. Data represent fold change normalized to KHOS-shGFP control cells transfected with pcDNA3 plasmid ± s.d. (N = 3; *Po0.05). (f) SATB2 model for OS invasion. In OS, high SATB2 regulates expression of genes to increase adhesion and decrease stress fibers, resulting in increased OS invasion and migration. SATB2 downregulates the expression of its target EPLIN and inhibits invasion by increasing and decreasing phosphorylation of paxillin and adhesion, respectively. In OS, SATB2 suppresses the inhibitory effect of EPLIN to increase invasion. Modulation of SATB2 levels also affects Rho/Rac, which have previously been reported to be regulated by EPLIN. which SATB2 is upregulated in OS has not been determined, both overexpression and shRNA led to increased and decreased miR-31 and 34b/c have been shown to target SATB2 transcripts invasion, respectively, suggest that, like SATB1 in breast cancer, and these miRNA clusters have been implicated in metastases and SATB2 has pro-metastatic functions in OS. Cancer cells exploit bone differentiation.8,25 Thus, further studies to elucidate SATB2 many different biological pathways to adapt to and survive during upstream regulation may provide insights into miRNA pathways different parts of the metastatic cascade such as invasion and involved in the progression of OS. colonization. Among many pathways, cytoskeleton-remodeling Our findings of increased SATB2 in murine OS lung metastases and focal adhesion pathways have previously been shown to be and demonstration that modulating SATB2 levels by critical in cancer metastasis.28 However, the molecular

© 2015 Macmillan Publishers Limited Oncogene (2015) 3582 – 3592 The role of SATB2 in osteosarcoma invasion BKA Seong et al 3590 mechanisms by which cytoskeletal-related genes are regulated in may regulate OS invasion by modulating cell adhesion and focal OS are largely unknown. Previous studies have reported that the adhesion turnover. Thus, our model suggests that SATB2 cytoskeleton linker ezrin provides an early survival advantage for promotes OS migration and invasion by reprogramming gene OS cells colonizing the lung,29 and CD99 suppresses OS cell expression to modulate and prevent hyperactivation of focal migration by inhibiting ROCK2,30 a critical effector of RhoA adhesion proteins and RhoA signaling. signaling. In our study, we demonstrated that SATB2 promotes OS Taken together, our findings demonstrate that SATB2 is highly invasion by, in part, by downregulating the expression of the expressed in OS cells, is a specific marker to distinguish OS from cytoskeleton-associated actin-binding protein EPLIN. Inhibition of other sarcoma tumors, and that SATB2 modulates the actin EPLIN in cells without SATB2 partially rescued the decreased cytoskeleton and focal adhesion pathways to promote OS invasion phenotype of sh-SATB2 OS cells. Interestingly, EPLIN invasion, in part via its regulation of EPLIN. Although actin- expression has been previously inversely correlated with cancer targeting drugs have been associated with significant toxicities in metastasis. For example, downregulation of EPLIN has been pre-clinical models due to the lack of discrimination between actin reported to be associated with epithelial–mesenchymal transition expressed in tumor cells and normal cells, recent studies suggest and was correlated with lymph-node metastasis in prostate that specific pharmacologic targeting of actin regulatory proteins cancer,13 however, the mechanisms by which EPLIN expression that are differentially expressed in tumor (vs normal) cells may be is regulated were not identified. Our study suggests that SATB2 a more feasible approach that will have less toxicity.28,42 Thus, regulation of EPLIN may occur in prostate and other non-OS understanding how SATB2 regulates transcription of genes and tumors. Although we did not detect SATB2 binding sites in the signaling pathways, including EPLIN, which govern actin cytoske- EPLIN promoter, a recent publication by Steder et al.31 provides a leton effects on motility and invasion may lead to the discovery of possible mechanism by which SATB2 may indirectly regulate proteins that can be targeted in metastatic OS and other cancers EPLIN. These investigators showed that the truncated p73 isoform with high expression of SATB2. ΔNp73 directly inhibits EPLIN expression by binding to the EPLIN promoter and that this leads to enhanced invasion of melanoma cells. We have previously reported that SATB2 binds to the closely MATERIALS AND METHODS related ΔNp63 and in HNSCC, augmenting ΔNp63 binding to p53- Antibodies and immunoblot analyses family consensus promoter sites in target genes. Thus, similar to Cells were lysed in EBC buffer (50 mM Tris (pH 8.0), 120 mM NaCl, 0.5% the role of SATB2 binding to ΔNp63 in HNSCCs, it is possible that Nonidet P-40) supplemented with complete protease inhibitors (Roche, SATB2 modulates the activity of p53 family proteins to repress Indianapolis, IN, USA) and protein concentration determined by Bradford EPLIN expression in OS. method (Bio-Rad, Hercules, CA, USA). In addition to EPLIN, we demonstrated that SATB2 regulates the For immunoblots, proteins were resolved by SDS-polyacrylamide gel expression of additional genes associated with cytoskeletal electrophoresis, transferred onto nitrocellulose (Bio-Rad), blocked in TBS dynamics and adhesion such as ABI3, GSN and FMN2. The (10 mM Tris (pH 8.0), 150 mM NaCl) with 5% (wt/vol) milk/bovine serum albumin and incubated with primary antibodies at 4 °C followed by pathways altered by SATB2 are similar to those identified in 6 horseradish peroxidase-conjugated secondary antibodies. Bound antibo- breast cancer cells lacking SATB1. This is likely due to conserved dies were detected by chemiluminenscence with SuperSignal West Pico DNA-binding consensus sites in target genes that are recognized (Thermo Scientific, Rockford, IL, USA). A list of antibodies is included in the by SATB family proteins. Moreover, our finding is also in Supplementary Information. agreement with previous gene expression profiling studies using both in vitro and in vivo models that identified both cytoskeleton- – Migration and invasion assays remodeling and focal adhesion as metastatic pathways in OS.32 35 Migration was analyzed using Boyden Chambers assay with 8-μm pore Consistent with our pathway analyses, OS cells with SATB2 inserts (BD Biosciences, Mississauga, ON, Canada). Cells in 0.1% FBS growth knockdown were characterized by morphological differences in media (100 μl) were seeded into the upper chamber and the lower the actin cytoskeleton. Sh-SATB2 cells were more elongated with chamber contained 10% FBS containing growth media. For invasion, 6% increased stress fiber formation. Biochemical experiments showed matrigel (BD Biosciences) in serum-free media was added to the upper increased RhoA activation, and decreased Rac1 activation chamber. Following 16–24 h incubation (37 °C, 5% CO2), cells that migrated suggesting that the highly regulated balance between RhoA and or invaded into the bottom chamber were stained with crystal violet and Rac1 signaling is altered, resulting in decreased migration of detected by Northern Eclipse, Empix Imaging Inc (Missisauga, ON, Canada) 43 sh-SATB2 cells. The activity of RhoA in cell adhesion exhibits a and ImageJ analyses of images of five random fields of view per biphasic relationship whereby RhoA activity is inhibited during chamber. Values were normalized to control, and expressed as fold initial cell adhesion and spreading. In the later phases of cell changes with standard deviations of triplicates. For scratch-wound assays, 36 cells were grown to confluent monolayers. A line was scratched through adhesion RhoA activity gradually increases. Our observations are cells in three replicates and low-serum media (DMEM/0.5% FBS) was in agreement with this idea in that sh-SATB2 cells, which have added. Images were captured at 0, 6, 12 and 24 h, and wound widths were increased RhoA activity, exhibit decreased adhesion in early measured, normalized to control, and expressed as percent recovery. phases of spreading (within 30 min) and increased phosphoryla- Experiments were performed ⩾ 3 times unless otherwise indicated in tion of FAK and paxillin at later phases (416 h post plating). legends. Considering the role of phosphorylated FAK and paxillin in focal adhesion turnover, the increase in the phosphorylation of FAK and shRNA knockdown and transfections paxillin suggests that sh-SATB2 cells may have altered turnover Viral work was performed in accordance with the Hospital for Sick Children rate of focal adhesions compared with control cells that express (Toronto, ON, Canada) safety guidelines and lentivirus constructs and virus high levels of SATB2. production were described previously.7 Details of shRNA and siRNA Previous reports demonstrate that EPLIN regulates cell knockdown and transfection experiments and target sequences are adhesion; however, the molecular mechanism is poorly provided in Supplementary Information. understood.37,38 We have now connected this important role for EPLIN to its regulation by SATB2 demonstrating that EPLIN rescues Microarray analysis and RT-PCR the decreased adhesion, invasion and increased stress fiber Total RNA was extracted with TRIZOL (Invitrogen, Carlsbad, CA, USA) from phenotypes detected in sh-SATB2 cells. Furthermore, we report KHOS-shSATB2 and control cells. Analysis of GeneChip Human Gene 1.0 ST that EPLIN regulates the level and phosphorylation of paxillin. This Arrays (Affymetrix, Santa Clara, CA, USA) was performed by the Microarray novel finding together with known roles for phosphorylated Analysis and Gene Expression Facility at The Hospital for Sick Children. – paxillin affecting focal adhesion turnover39 41 suggests that EPLIN Pathway analyses were conducted using the Partek software (KEGG based)

Oncogene (2015) 3582 – 3592 © 2015 Macmillan Publishers Limited The role of SATB2 in osteosarcoma invasion BKA Seong et al 3591 and GSEA tool (Broad Institute) with a P-value cutoff of 0.05. The cells and correlates with clinical lymph node metastasis. Oncogene 2011; 30: differentially regulated pathway networks were visualized using Enrich- 4941–4952. ment Map Plug-in in Cytoscape44 with P-value and Q-value cutoff of 0.05 14 Maul RS, Chang DD. EPLIN, epithelial protein lost in neoplasm. Oncogene 1999; 18: and 0.1, respectively. RT-PCR conditions and primers are included in 7838–7841. methods section in Supplementary Information. 15 Abe K, Takeichi M. EPLIN mediates linkage of the cadherin catenin complex to F-actin and stabilizes the circumferential actin belt. Proc Natl Acad Sci USA [Internet] 2007; 105: – fl 13 19. Available from http://eutils.ncbi.nlm.nih.gov/entrez/eutils/elink.fcgi?dbfrom = GTPase activation, adhesion and immuno uorescence assays pubmed&id = 18093941&retmode = ref&cmd = prlinks. For GTPase activation assay, cells were seeded (30% confluence), serum- 16 Molyneux SD, Di Grappa MA, Beristain AG, McKee TD, Wai DH, Paderova J et al. starved overnight and induced with 10% FBS-containing growth media for Prkar1a is an osteosarcoma tumor suppressor that defines a molecular subclass the indicated durations. Pull-downs using rhotekin-RBD and PAK1-PBD in mice. J Clin Invest 2010; 120: 3310–3325. beads were performed according to the manufacturer’s instructions 17 Latini FRM, Hemerly JP, Oler G, Riggins GJ, Cerutti JM. Re-expression of (Cytoskeleton, Inc, Denver, CO, USA). For adhesion assays, cells were ABI3-binding protein suppresses thyroid tumor growth by promoting senescence seeded in triplicate in 96-well plates and allowed to adhere and spread. At and inhibiting invasion. Endocr Relat Cancer 2008; 15:787–799. indicated time points, non-adherent cells were washed and adherent cells 18 Yuan X, Yu L, Li J, Xie G, Rong T, Zhang L et al. ATF3 suppresses metastasis were fixed with 4% paraformaldehyde, washed, stained with crystal violet, of bladder cancer by regulating gelsolin-mediated remodeling of the actin 73 – and the absorbance at 570 nm was read. Experiments were performed ⩾ 3 cytoskeleton. Cancer Res 2013; :3625 3637. times unless otherwise indicated in legends. For immunofluorescence, cells 19 Zech T, Calaminus SDJ, Caswell P, Spence HJ, Carnell M, Insall RH et al. The Arp2/3 were seeded on fibronectin-coated culture slides, fixed with 4% activator WASH regulates 5 1-integrin-mediated invasive migration. J Cell Sci 2011; 124 – paraformaldehyde, and permeabilized with 0.2% Triton X-100. Slides were : 3753 3759. 20 Sahai E, Marshall CJ. RHO-GTPases and cancer. Nat Rev Cancer 2002; 2:133–142. blocked with 6% normal goat serum and 0.5% bovine serum albumin 21 Maul RS. EPLIN regulates actin dynamics by cross-linking and stabilizing filaments. (BSA), incubated with the indicated primary and secondary antibodies. J Cell Biol 2003; 160:399–407. Images were taken with Olympus IX81 Quorum Spinning Disk Confocal 22 Satoh Y, Yokota T, Sudo T, Kondo M, Lai A, Kincade PW et al. The Satb1 protein Microscope (Guelph, ON, Canada) and analyzed with the Perkin Elmer directs hematopoietic stem cell differentiation toward lymphoid lineages. Volocity software (PerkinElmer, Woodbridge, ON, Canada). Immunity 2013; 38:1105–1115. 23 Zhao X, Qu Z, Tickner J, Xu J, Dai K, Zhang X. The role of SATB2 in skeletogenesis and human disease. Cytokine Growth Factor Rev 2014; 25:35–44. CONFLICT OF INTEREST 24 Leoyklang P, Suphapeetiporn K, Siriwan P, Desudchit T, Chaowanapanja P, The authors declare no conflict of interest. Gahl WA et al. Heterozygous nonsense mutation SATB2 associated with cleft palate, osteoporosis, and cognitive defects. Hum Mutat 2007; 28:732–738. 25 Wei J, Shi Y, Zheng L, Zhou B, Inose H, Wang J et al. miR-34s inhibit osteoblast ACKNOWLEDGEMENTS proliferation and differentiation in the mouse by targeting SATB2. J Cell Biol 2012; 197 – This study was supported by funds from the Canadian Cancer Society Research :509 521. Institute (#2010-700580, MSI), Ontario Research Fund (MSI, DM) and a Canada 26 Hassan MQ, Gordon JAR, Beloti MM, Croce CM, van Wijnen AJ, Stein JL et al. A network connecting Runx2, SATB2, and the miR-23a ~ 27a ~ 24-2 cluster reg- Research Chair in Cancer Biology (MSI). BKAS and JL received scholarships from ulates the osteoblast differentiation program. Proc Natl Acad Sci USA 2010; 107: Hospital for Sick Children Research Training Centre (Restracomp). We are grateful for 19879–19884. assistance with microarray analyses from Daniel Picard and Dr Annie Huang. 27 Conner JR, Hornick JL. SATB2 is a novel marker of osteoblastic differentiation in bone and soft tissue tumours. Histopathology 2013; 63:36–49. 28 – REFERENCES 28 Hall A. The cytoskeleton and cancer. Cancer Metastasis Rev 2009; (1-2): 5 14. 29 Khanna C, Wan X, Bose S, Cassaday R, Olomu O, Mendoza A et al. The membrane- 1 Broadhead ML, Clark JCM, Myers DE, Dass CR, Choong PFM. The molecular cytoskeleton linker ezrin is necessary for osteosarcoma metastasis. Nat Med 2004; pathogenesis of osteosarcoma: a review. Sarcoma 2011; 2011:1–12. 10:182–186. 2 Britanova O, Depew MJ, Schwark M, Thomas BL, Miletich I, Sharpe P et al. Satb2 30 Zucchini C, Manara MC, Pinca RS, De Sanctis P, Guerzoni C, Sciandra M et al. CD99 haploinsufficiency phenocopies 2q32-q33 deletions, whereas loss suggests a suppresses osteosarcoma cell migration through inhibition of ROCK2 activity. fundamental role in the coordination of jaw development. Am J Hum Genet 2006; Oncogene 2014; 33:1912–1921. 79: 668–678. 31 Steder M, Alla V, Meier C, Spitschak A, Pahnke J, Fürst K et al. DNp73 exerts 3 Dobreva G, Chahrour M, Dautzenberg M, Chirivella L, Kanzler B, Fariñas I et al. function in metastasis initiation by disconnecting the inhibitory role of EPLIN on SATB2 is a multifunctional determinant of craniofacial patterning and osteoblast IGF1R-AKT/STAT3 signaling. Cancer Cell 2013; 24:512–527. differentiation. Cell 2006; 125:971–986. 32 Khanna C, Khan J, Nguyen P, Prehn J, Caylor J, Yeung C et al. Metastasis- 4 Dobreva G. SUMO modification of a novel MAR-binding protein, SATB2, associated differences in gene expression in a murine model of osteosarcoma. modulates immunoglobulin gene expression. Genes Dev 2003; 17: 3048–3061. Cancer Res 2001; 61: 3750–3759. 5 Britanova O, de Juan Romero C, Cheung A, Kwan KY, Schwark M, Gyorgy A et al. 33 Flores RJ, Li Y, Yu A, Shen J, Rao PH, Lau SS et al. A systems biology approach Satb2 is a postmitotic determinant for upper-layer neuron specification in the reveals common metastatic pathways in osteosarcoma. BMC Syst Biol 2012; 6:50. neocortex. Neuron 2008; 57: 378–392. 34 O'Donoghue LE, Ptitsyn AA, Kamstock DA, Siebert J, Thomas RS, Duval DL. 6 Han H-J, Russo J, Kohwi Y, Kohwi-Shigematsu T. SATB1 reprogrammes gene Expression profiling in canine osteosarcoma: identification of biomarkers and expression to promote breast tumour growth and metastasis. Nature 2008; 452: pathways associated with outcome. BMC Cancer 2010; 10: 506. 187–193. 35 Jones KB, Salah Z, Del Mare S, Galasso M, Gaudio E, Nuovo GJ et al. miRNA 7 Chung J, Lau J, Cheng LS, Grant RI, Robinson F, Ketela T et al. SATB2 augments signatures associate with pathogenesis and progression of osteosarcoma. Cancer ΔNp63α in head and neck squamous cell carcinoma. EMBO Rep 2010; 11: Res 2012; 72: 1865–1877. 777–783. 36 Huveneers S, Danen EHJ. Adhesion signaling—crosstalk between integrins, 8 Aprelikova O, Yu X, Palla J, Wei B-R, John S, Yi M et al. The role of miR-31 and its Src and Rho. J Cell Sci 2009; 122:1059–1069. target gene SATB2 in cancer-associated fibroblasts. Cell Cycle 2010; 9: 4387–4398. 37 Tsurumi H, Harita Y, Kurihara H, Kosako H, Hayashi K, Matsunaga A et al. Epithelial 9 Patani N, Jiang W, Mansel R, Newbold R, Mokbel K. The mRNA expression of protein lost in neoplasm modulates platelet-derived growth factor-mediated SATB1 and SATB2 in human breast cancer. Cancer Cell Int 2009; 9:18. adhesion and motility of mesangial cells. Kidney Int 2014; 86:548–557. 10 Eberhard J, Gaber A, Wangefjord S, Nodin B, Uhlén M, Ericson Lindquist K et al. 38 Sanders AJ, Ye L, Mason MD, Jiang WG. The impact of EPLINα (Epithelial protein A cohort study of the prognostic and treatment predictive value of SATB2 lost in neoplasm) on endothelial cells, angiogenesis and tumorigenesis. expression in colorectal cancer. Br J Cancer 2012; 106: 931–938. Angiogenesis 2010; 13: 317–326. 11 Olson MF, Sahai E. The actin cytoskeleton in cancer cell motility. Clin Exp 39 Quizi JL, Baron K, Al-Zahrani KN, Reilly POA, Sriram RK, Conway J et al. Metastasis 2008; 26: 273–287. SLK-mediated phosphorylation of paxillin is required for focal adhesion turnover 12 Parri M, Chiarugi P. Rac and Rho GTPases in cancer cell motility control. Cell and cell migration. Oncogene 2012; 32: 4656–4663. Commun Signal 2010; 8: 23. 40 Zaidel-Bar R, Milo R, Kam Z, Geiger B. A paxillin tyrosine phosphorylation switch 13 Zhang S, Wang X, Osunkoya AO, Iqbal S, Wang Y, Chen Z et al. EPLIN regulates the assembly and form of cell-matrix adhesions. J Cell Sci 2006; 120: downregulation promotes epithelial–mesenchymal transition in prostate cancer 137–148.

© 2015 Macmillan Publishers Limited Oncogene (2015) 3582 – 3592 The role of SATB2 in osteosarcoma invasion BKA Seong et al 3592 41 Abou Zeid N, Vallés A-M, Boyer B. Serine phosphorylation regulates paxillin 43 Schneider CA, Rasband WS, Eliceiri KW. NIH Image to ImageJ: 25 years of image turnover during cell migration. Cell Commun Signal 2006; 4:8. analysis. Nat Methods 2012; 9: 671–675. 42 Stehn JR, Haass NK, Bonello T, Desouza M, Kottyan G, Treutlein H et al. A novel 44 Merico D, Isserlin R, Stueker O, Emili A, Bader GD. Enrichment map: a network- class of anticancer compounds targets the actin cytoskeleton in tumor cells. based method for gene-set enrichment visualization and interpretation. PLoS ONE Cancer Res 2013; 73: 5169–5182. 2010; 5: e13984.

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