Oncogene (2013), 1–10 & 2013 Macmillan Publishers Limited All rights reserved 0950-9232/13 www.nature.com/onc

ORIGINAL ARTICLE

Hedgehog signaling induces osteosarcoma development through Yap1 and H19 overexpression LH Chan1, W Wang1, W Yeung1, Y Deng1, P Yuan2 and KK Mak1,3,4 Osteosarcoma is one of the most common bone tumors. However, the genetic basis for its pathogenesis remains elusive. Here, we investigated the roles of Hedgehog (Hh) signaling in osteosarcoma development. Genetically-engineered mice with ubiquitous upregulated Hh signaling specifically in mature osteoblasts develop focal bone overgrowth, which greatly resembles the early stage of osteosarcoma. However, these mice die within three months, which prohibits further analysis of tumor progression. We therefore generated a mouse model with partial upregulated Hh signaling in mature osteoblasts and crossed it into a p53 heterozygous background to potentiate tumor development. We found that these mutant mice developed malignant osteosarcoma with high penetrance. Isolated primary tumor cells were mainly osteoblastic and highly proliferative with many characteristics of human osteosarcomas. Allograft transplantation into immunocompromised mice displayed high tumorigenic potential. More importantly, both human and mouse tumor tissues express high level of yes-associated protein 1 (Yap1), a potent oncogene that is amplified in various cancers. We show that inhibition of Hh signaling reduces Yap1 expression and knockdown of Yap1 significantly inhibits tumor progression. Moreover, long non-coding RNA H19 is aberrantly expressed and induced by upregulated Hh signaling and Yap1 overexpression. Our results demonstrate that aberrant Hh signaling in mature osteoblasts is responsible for the pathogenesis of osteoblastic osteosarcoma through Yap1 and H19 overexpression. Oncogene advance online publication, 21 October 2013; doi:10.1038/onc.2013.433 Keywords: osteosarcoma; yes-associated protein1; long non-coding RNA H19

INTRODUCTION Osteosarcoma is one of the most common malignant bone tumors in children and adolescents.1,2 The five-year survival rate for osteosarcoma patients is around 70%, but for patients with recurrent disease or metastasis at diagnosis this drops significantly to only 20–30%. To date, genetic contributions to the pathogenesis of osteosarcoma remain unclear and the molecular bases for the development of different osteosarcoma subtypes remain elusive. Previous genetic mouse models of osteosarcoma are mostly fibroblastic or undifferentiated in nature3–5 but the osteoblastic form is lacking. A recent report indicates that conditional knockout of the p53 allele versus knockdown of p53 using shRNA approach resulted in different subtypes of osteosarcoma.6 This raises the importance of the signaling intensity and timing of genetic manipulation at various stages within the osteoblastic lineage that contributes to different subtypes of the bone tumor. More research in this area will advance further development of better therapeutic interventions for precise treatment of each osteosarcoma subtype. Hedgehog (Hh) signaling has been widely implicated in the development of multiple cancers in which many of them can be largely phenocopied by mutant mouse models.7–12 For instance, aberrant activation of Hh signaling is implicated in human medulloblastoma. It has been shown that transcriptional cofactor yes-associated protein 1 (Yap1), a downstream effector of tumor suppressive Hippo pathway, is responsible for a subset of Hh-

associated medulloblastoma.13 In addition, YAP1 overexpression is frequently found in different human cancers and many of them are related to a dysregulated Hippo signaling pathway.14–18 In the context of osteoblast lineage, Yap1 interacts with Runx2, an important transcription factor for osteoblast commitment, to regulate Osteocalcin expression.19,20 However, whether Hh signaling and Hippo signaling are implicated in the pathogenesis of osteosarcoma remained elusive. Previous studies showed that Hh signaling may be correlated with osteosarcoma.21,22 Inhibition of Hh signaling or GLI2 knockdown in human osteosarcoma cell lines reduces cell proliferation and enhances apoptosis. However, these results are mainly based on in vitro analyses and expression association studies in patient specimens without detailed investigation of the molecular mechanisms. Long non-coding RNAs (lncRNAs) play a key regulatory role in the development of various tumors. Aberrant expression of certain lncRNAs causes them to function as oncogenes, and they are associated with tumor growth and invasion.23–27 Among these, H19 is one of the most extensively studied candidates. H19 is an imprinted gene expressed from the maternal allele during embryonic development and repressed after birth.28 Loss of imprinting of H19 in CpG methylation of CCCTC-binding factor (CTCF)-binding site is associated with osteosarcoma.29 However, the molecular mechanism in regulating H19 expression remains unclear. Here, we have addressed the importance of Hh signaling in the onset of osteoblastic osteosarcoma using geneticallyengineered mouse models. We generated an osteoblastic

1 Key Laboratories for Regenerative Medicine, Ministry of Education, School of Biomedical Sciences, Faculty of Medicine, The Chinese University of Hong Kong, Shatin, Hong Kong SAR; 2Department of Chemical Pathology, The Chinese University of Hong Kong, Hong Kong SAR; 3Stem Cell and Regeneration Thematic Research Program, School of Biomedical Sciences, The Chinese University of Hong Kong, Shatin, Hong Kong SAR and 4CUHK Shenzhen Research Institute, The Chinese University of Hong Kong, Shenzhen, China. Correspondence: Dr KK Mak, Stem Cell and Regeneration Thematic Research Program, School of Biomedical Sciences, The Chinese University of Hong Kong, Shatin, Hong Kong. E-mail: [email protected] Received 24 July 2013; revised 6 September 2013; accepted 13 September 2013

Hh induced Yap1/H19 overexpression in osteosarcoma LH Chan et al

2 osteosarcoma mouse model and found that upregulated Hh signaling induces osteosarcoma through Yap1 and H19 overexpression. In addition, Yap1 is found to be highly expressed in both human and mouse osteosarcomas. Our data indicate that aberrant Hh signaling in mature osteoblasts leads to the development of osteoblastic osteosarcoma. RESULTS Upregulated Hh signaling induces osteoblastic osteosarcoma We previously generated the Ptch1c/c;HOC-Cre mutant mice in which Hh signaling is upregulated specifically in mature osteoblasts.30 Interestingly, apart from the bone remodeling defects, focal bone overgrowth was frequently observed in long bones of the Ptch1c/c; HOC-Cre mutant mice (Supplementary Figure S1a–c). Serum alkaline phosphatase level was significantly increased.30 These observations are very similar to the early stage of osteosarcoma development and suggest a role of Hh signaling in the pathogenesis of this disease. Unfortunately, the mutant mice died within three months due to unknown reasons, which prohibited a detailed investigation of tumor progression. In order to further examine the functional roles of Hh signaling in osteosarcoma development, we partially upregulated Hh signaling by using a heterozygous mutant, Ptch1c/ þ ;HOC-Cre, in which only one copy of the Ptch1 allele is removed in mature osteoblasts. The survival rate of the heterozygous mice was comparable to that of the wild-type mice without any bone tumor formation within the 16 months observation (Figure 1a). In order to enhance the frequency of osteosarcoma occurrence, we crossed the Ptch1c/ þ ;HOC-Cre mutant mice to a p53  /  mutant mice to generate Ptch1c/ þ ;p53 þ /  ; HOC-Cre mutant mice. p53  /  mutant mice have been shown to develop multiple cancers including osteosarcoma, but the

occurrence of osteosarcoma is very low by itself31 because the effect is masked by the early onset of other cancers. The Ptch1c/ þ ; p53 þ /  ;HOC-Cre mutant mouse is viable, fertile and developmentally normal as compared to age-matched wild-type mice. Interestingly, these mice started to develop observable bone tumors as early as 7 months old, and the incidence significantly increased at 11 months onwards with about 70% penetrance (Figure 1a). These tumors were located mostly in the forelimbs and hind limbs (Figures 1b and d), which are the most common primary sites in human osteosarcoma. More interestingly, bone tumors were also frequently found in the spine (Supplementary Figure S1d), which led to lower body paralysis of the mutant mice. Bone tumors in other locations such as ribs and skull were less frequently found (Supplementary Figure S1e and f). Occasionally, pulmonary metastasis was noted in the lungs with tumor nodule formation (Supplementary Figure S1g). X-ray analysis revealed that the primary tumors were composed of highly mineralized tissues regardless of the site of occurrence (Figure 1c). Histological sections confirmed that the tumors were composed of abundant osteoids with multinucleated cells (Figure 1e). All these phenotypes greatly resemble the characteristics of human osteoblastic osteosarcoma. Molecular analysis showed that Hh targets Gli1, Gli2 and Hhip1 were all highly upregulated in the tumor tissues (Figure 1f). Osteoblast markers Runx2, Osteocalcin (Oc) and Osteopontin (Opn) were significantly increased (Figure 1g). These data suggest that the origins of the tumors were derived from osteoblasts with upregulated Hh signaling. In addition, the majority of the Ptch1c/ þ ;p53 þ /  ;HOC-Cre mutant mice developed only bone tumors, and not tumors in other organs within the observation period. Only a few p53 þ /  mutant mice (n ¼ 3) showed spontaneous tumor development in other organs, which is not relevant to this study (Figure 1a). Altogether, these data suggest that aberrant Hh signaling in mature osteoblasts induces osteoblastic osteosarcoma.

Figure 1. Upregulated Hh signaling contributes to osteosarcoma development. (a) Kaplan–Meier survival plot for mouse with the indicated genotypes: p53 þ /  (n ¼ 20); Ptch1c/ þ ;HOC-Cre (n ¼ 24); Ptch1c/ þ ;p53 þ /  ;HOC-Cre (22 out of 31 developed osteosarcoma within the 16 months period; 70% penetrance); Ptch1c/c;HOC-Cre;p53 þ /  (n ¼ 3). Representative presentation of osteosarcoma in forelimb (b) at 7 months old and hind limb (d) at 11 months old, respectively. (c) X-ray analysis of the tumor mass from (b). White arrow points to area with intensive mineralization of the tumor. (e) Representative histological section from (d) with massive osteoids. Scale bar, 200mm. Inbox showed a higher magnification of the tumor tissues with osteoblastic nature. (f and g) Quantitative RT-PCR of tumor tissues isolated from the Ptch1c/ þ ; p53 þ /  ;HOC-Cre for Hh target genes and specific bone markers, respectively (n ¼ 3). Statistical analyses are presented by student t-test as shown (***Po0.01). Oncogene (2013) 1 – 10

& 2013 Macmillan Publishers Limited

Hh induced Yap1/H19 overexpression in osteosarcoma LH Chan et al

3

WT calvarial osteoblasts Kios-5 ***

8 6

** ***

4 2 0 Runx2

Opn

Oc

WT calvaria

6

were more PH3-positive cells as compared to that of wild-type calvarial osteoblasts or MC3T3-E1 cells, respectively (Figures 2b and c and Supplementary Figure S2b and c). In addition, MTT assay also revealed a significant increase in cell proliferation rate (Supplementary Figure S2d). When the osteoblasts were cultured in osteogenic condition, Kios cells showed a significant increase of mineralized bone nodules with Alizarin Red and von Kossa staining when compared with control osteoblasts (Figure 2d and Supplementary Figure S2e). These data suggest that osteoblast differentiation and maturation in the primary Kios cells were highly accelerated. Other typical features of tumorigenicity are enhanced cell migration and invasion. Therefore, we examined the primary tumor cells with wound healing and trans-well invasion assay respectively. As expected, Kios cells showed a significant increase in cell migration and invasion as compared to that of the controls (Figures 2e and f and Supplementary Figure S2f). Furthermore, soft agar assay showed that Kios cells were able to form colonies, confirming their capability of anchorage-independent growth (Supplementary Figure S2g). These findings strongly indicate that the primary cells undergo malignant transformation that resembles features of cancer cells. Next, we tested the tumorigenicity of

*** ***

4 2 0 c-Fos

Proliferation Index

0.25 0.20

***

15 10

**

5 0 Survivin

Ezrin

Kios-5

cIAP

WT Calvarial osteoblasts Kios-5 **

0.15 0.10 0.05 0.00

0 hr

Kios-1

20

10 hr

14 hr

Kios-1

von Kossa

Alizarin Red

WT calvarial osteoblasts

WT calvarial osteoblasts

Relative abundance

10

Relative abundance

Relative abundance

Primary tumor cells possess cancer characteristics both in vitro and in vivo In order to further analyze the cancer properties of the tumors from the Ptch1c/ þ ;p53 þ /  ;HOC-Cre mutant mice, we isolated primary cells from the tumors and cultured them for cellular and molecular analysis. First, we tested the origin of the tumor cells by making use of the LacZ knockin allele derived from the floxed allele of Ptch1 in which Cre-mediated recombination induces LacZ expression.32 We assessed the primary tumor cells named Kios-5 in culture by X-gal staining. As expected, strong expression was detected from the isolated cells (Supplementary Figure S2a). They also expressed high levels of osteoblast markers as well as cancer marker genes such as c-Fos, Ezrin, Survivin and cIAP when compared with wild-type calvarial osteoblasts (Figure 2a). We isolated primary cells from at least seven individual tumors (Kios59S1) from different Ptch1c/ þ ;p53 þ /  ;HOC-Cre mutant mice, and they all showed consistent results. These findings suggest that the isolated primary cells were osteosarcoma cancer-like cells. Since many of the cancer cells are highly proliferative, we tested the proliferation of the primary cells by examining the expression of phospho-histone H3 (PH3). In the Kios primary tumor cells, there

WT calvarial osteoblasts

Kios-5

Figure 2. Cancer properties of primary osteosarcoma cells isolated from the Ptch1c/ þ ;p53 þ /  ;HOC-Cre mutant mice. (a) Quantitative RT-PCR of genes implicated in osteoblast differentiation and cancer progression. (b) Cell proliferation of primary osteosarcoma cells by PH3 staining. More PH3-positive cells were found in the Kios-5 cells. Scale bar, 50 mm. (c) Statistical analysis of (b). (d) Kios-1 cells were cultured in an osteogenic medium and stained with Alizarin Red at Day 6 and von Kossa at Day 16, respectively. Enhanced mineralization was observed. (e) Wound healing assay of primary osteosarcoma cells. Cells were pretreated with mitomycin C (4 ug/ml) for 2 h before the assays. Kios-1 cells showed enhanced cell migration with complete closure after 14 h. Dotted lines represent the cell fronts of the gaps. (f ) Transwell invasion assay in Matrigel-coated chambers. After 48 h incubation, cells underneath the chambers were stained with crystal violet. Kios-5 cells showed enhanced invasion. Statistical analyses are presented by student t-test as shown (***Po0.01; **Po0.05). & 2013 Macmillan Publishers Limited

Oncogene (2013) 1 – 10

Hh induced Yap1/H19 overexpression in osteosarcoma LH Chan et al

4 and Cyr61 were all strongly activated, respectively (Figures 4b and c and Supplementary Figure S4a and b). In addition, GDC0449 treatment successfully inhibited their expression at RNA (Figures 4b and c and Supplementary Figure S3c and d) and protein level (Figure 4d and Supplementary Figure S4d), respectively. Using a TEAD/Sd luciferase reporter,33 we found that the transcriptional activity of Yap1 was increased in the Kios-5 primary cells, and its activity was inhibited by GDC-0449 treatment (Figure 4e). Furthermore, Amphiregulin, a recently identified downstream target of Hippo signaling34 was also upregulated in both tumor tissues and primary cancer cells (Supplementary Figure S4e and f). Altogether, these data indicate that Hippo signaling is regulated by Hh signaling during osteosarcoma development. We also isolated primary osteoblasts directly from the calvaria of the neonatal Ptch1c/c; HOC-Cre mutant mice and found that Hippo effectors and target genes were highly upregulated (Supplementary Figure S4g and h). These data indicate that Hh signaling regulates the Hippo pathway independent of p53 signaling in our mouse models.

the primary cells in vivo by injecting primary tumor cells into nude mice (Figure 3). Both subcutaneous injection and intra-tibial injection formed bone tumors with over 80% incidence rate (Figures 3a and e). These data indicate that primary cells isolated from the tumors are highly tumorigenic and possess characteristics very similar to human osteosarcoma. Hh signaling induces Yap1 overexpression for osteosarcoma development To test whether malignant transformation was caused by aberrant Hh signaling, we tested the expression of Hh target genes in the primary Kios cells. Both Gli1 and Gli2 were highly upregulated in the Kios cells as compared to the controls (Figure 4a and Supplementary Figure S3a-b). Next, we treated the primary tumor cells with Hh-specific inhibitor GDC-0449. In the Kios cells, both Gli1 and Gli2 expression were significantly reduced with GDC-0449 treatment (Figures 4a and Supplementary Figure S3c and d). To further understand the molecular mechanism underlying osteosarcoma development, we searched signaling pathways that have been widely implicated in cancer development. We found that Hippo signaling effector Yap1 was highly expressed in the tumor tissues isolated from our Ptch1c/ þ ;p53 þ /  ;HOC-Cre mutant mice (Figure 4f). Similarly, primary Kios cells also showed strong Yap1 protein expression (Figure 4d and Supplementary Figure S4c and d) and nuclear localization as compared to that of the wild type (Figure 4g). From all the primary cells being isolated from different tumors, the expression of Yap1 and Taz, the downstream effectors of Hippo pathway, as well as the Hippo target genes Ctgf

Inhibition of Yap1 expression attenuates tumor formation To test whether Yap1 overexpression led to osteosarcoma tumor formation, we knocked down Yap1 expression in the primary tumor cells by ShYAP approach. To avoid non-specific binding effects, two independent ShYAP lentiviruses were tested and both of them reduced the proliferation of the primary tumor cells as shown by reduced PH3 expression (Figures 5a and b). More importantly, when these transduced cells were injected into nude

Kios5 Tumor incidence Tumor latency (Days)

Median Mean SD

Subcutaneous N=12/14 86% 18 21.9 19.7

Kios-1

Intra-tibia N=16/18 89% 23 26.2 10.6

Subcutaneous

Subcutaneous N=4/5 80% 21 20.8 1.7

Intra-tibia N=4/5 80% 25 27 28 26 9.3 3.2

Intra-tibial

80

60

60 Days

Days

40 40

20 20 0

0

Figure 3. In vivo cancer properties of primary osteosarcoma cells. Kios-5 cells were injected into nude mice through subcutaneous (n ¼ 14) (a) or intra-tibial (orthotopic) route (n ¼ 18) (c). Black arrows point to the site of injection. Tumors were isolated after Day 40 (b) and Day 28, (d) respectively. (e) Statistics of the incidence and latency of tumor development by both routes in Kios-5 and Kios-1 cells, respectively. Scatter plots of subcutaneous injection (f) and intra-tibial injection (g) of Kios-5 cells are shown. Oncogene (2013) 1 – 10

& 2013 Macmillan Publishers Limited

Hh induced Yap1/H19 overexpression in osteosarcoma LH Chan et al

5 40

***

*** 25

***

30

***

***

***

25

25 20 4

***

3 2

**

Relative abundance

30 Relative abundance

Relative abundance

35

20 15 10

***

***

5

1 0

0 Gli1

Gli2

20 15 3

Kios-5

1

Ctgf

Taz

MC3T3-E1 GDC-0449 +

***

0 Yap1

Kios-5

***

Cyr61 ***

pGL3 control

30

GDC-0449 +

pGL3 + 50nM GDC-0449 SD-luc

RLU

GDC-0449 +

***

2

40 WT calvaria

WT calvaria, WT calvaria, 50nM GDC-0449 Kios-5, Kios-5, 50nM GDC-0449

Yap1

SD-luc + 50nM GDC-0449

20

10

β-Actin

0 Ptch1c/+;p53 +/-;HOC-Cre

WT calvaria

Kios-5

DAPI

Yap1

WT

Figure 4. Yap1 overexpression in osteosarcoma. (a–c) Quantitative RT-PCR of gene expression in Kios-5 primary osteosarcoma cells as shown. Hh and Hippo effectors were both highly upregulated. (d) Western blot analysis of Kios-5 primary cells with Yap1 antibodies. b-Actin was used as control of total protein. Yap1 (65 kDa) overexpression was significantly reduced with Hh inhibitor GDC-0449 treatment. (e) Sd luciferase activity was increased in Kios-5 primary cells. GDC-0449 treatment reduced its activities after 48 h. pGL3 vector was used as a control. Statistical analyses are presented by student t-test as shown (***Po0.01; **Po0.05). (f ) Immunohistochemistry of Yap1 expression in osteosarcoma isolated from the mutant mice. Scale bar, 200 mm. Inbox showed higher magnification of the region as shown. (g) Immunocytochemistry of Yap1 expression in Kios-5 cells. Majority of Yap1 was localized within the nucleus. Scale bar, 10 mm.

mice subcutaneously, the tumor sizes were significantly smaller as compared to that of the control (Figure 5c). The expressions of Yap1 and downstream target genes were also inhibited (Figures 5d and e and Supplementary Figure S5a and b). These data indicate that Yap1 contributes to the aberrant Hh signaling in promoting tumor progression. Next, we tested if Yap1 overexpression is also implicated in human osteosarcoma. We examined the expression level of Yap1 in a human osteosarcoma tissue array and 78% of the patient specimens tested showed moderate to high levels of Yap1 expression (Figure 5f and Supplementary Table S1). When we compared the Yap1 expression level with the patient survival data (Supplementary Table S2), no significant correlation was observed. However, it is noted that higher number of osteosarcoma tissues with fibroblastic or chondroblastic origins showed either no or lower level of Yap1 expression. In addition, we also evaluated the correlation of Hh pathway activation and Yap1 expression in human osteosarcoma patient samples by using expression dataset from Jones et al.35 (Figure 5g). The expression of GLI1, GLI2 and Yap1 were all & 2013 Macmillan Publishers Limited

significantly upregulated in the osteosarcoma patient samples as compared to that of the normal samples. Altogether, these data suggest that Hippo signaling has an important role in the pathogenesis and progression of human osteosarcoma. H19 is induced by Yap1 overexpression To further identify the underlying molecular mechanism, we activated Yap1 by genetic removal of mammalian Hippo core components Mst1 and Mst2 simultaneously in embryonic stem cells and analyzed the gene expression profile changes by microarray analysis (Supplementary Figure S6). Our data showed that among many different genes that were highly upregulated by Yap1 activation, lncRNA H19 was one of the candidates with the highest relevance in cancer development (Supplementary Figure S6). We next tested H19 expression in the mouse osteosarcoma tissues and primary osteosarcoma cells. Consistent to the microarray analysis, H19 was highly upregulated in the tumor tissues and the primary Kios cells, respectively (Figures 6a and b). Oncogene (2013) 1 – 10

Hh induced Yap1/H19 overexpression in osteosarcoma LH Chan et al

6 ShYAP#3

0.25

ShYAP#4 Proliferation Index

ShLUC

0.20 0.15

1.5 1.0

0.5 ***

Normal human bone

ShYAP#4

ShLUC ShYAP#3 ShYAP#4

1.0

0.5 ***

***

***

***

0.0

0.0 ShLUC

ShYAP#3

ShYAP#4

Moderate YAP expression

Normal

Fold difference against internal control

***

Relative abundance

Relative abundance ShYAP#4

ShYAP#3

0.00

Yap1

1.5

ShYAP#3

**

0.05 ShLUC

ShLUC

**

0.10

Ctgf

Cyr61

Strong YAP expression

Osteosarcoma

***

***

***

Figure 5. Inhibition of Yap1 expression attenuates osteosarcoma progression. (a) Cell proliferation in primary osteosarcoma cells as shown by PH3 expression. Yap1 was subjected to knockdown by in vitro transduction of two independent ShYAP lentiviruses respectively as shown. Scale bar, 50 mm. A significant reduction in the number of PH3-positive cells was detected. (b) Statistical analysis of (a). (c) Kios-5 cells transduced with ShYAP#3 and ShYAP#4, respectively were injected subcutaneously into nude mice (n ¼ 5 per group). Black arrows point to the sites of injection. Tumor sizes were obviously smaller 21 days after injection (lower panel). (d and e) Quantitative RT-PCR of the expression of Yap1 and downstream target genes of Hippo signaling pathway were all inhibited in the tumor tissues from (c). (f) Examination of YAP1 expression in human osteosarcoma tissue array (SuperBioChips Laboratories) with 54 patient samples. Moderate (n ¼ 27) to strong (n ¼ 15) YAP1 expression was detected in most of the samples (78% of total specimen tested). Scale bar, 200 mm. Inbox showed higher magnification of the regions as shown. (g) Correlation of Hh pathway activation with YAP1 expression from a published human osteosarcoma expression dataset.35 Fourteen osteosarcoma patient samples were compared with four normal controls. Red arrows represent higher expression and blue arrows represent lower expression. Statistical analyses are presented by student t-test as shown (***Po0.01; **Po0.05).

Oncogene (2013) 1 – 10

& 2013 Macmillan Publishers Limited

Hh induced Yap1/H19 overexpression in osteosarcoma LH Chan et al

7

Figure 6. H19 expression in osteosarcoma development. Quantitative RT-PCR of H19 expression in osteosarcoma tissues (a) and primary tumor cells (b). H19 was strongly upregulated in both cases. (c) Knockdown of Yap1 expression in Kios-5 cells inhibited H19 expression. (d) Treatment with Hh-specific inhibitors GDC-0449 reduced H19 expression in Kios-5 cells after 24 h. (e) Forced expression of Yap1 induced H19 expression in MC3T3E1 osteoblast cell lines. Flag vector was used as control. (f ) Treatment with Shh (1 mg/ml) or Hh agonist purmorphamine (Pur; 1 mM) induced H19 expression in wild-type primary osteoblasts. Statistical analyses are presented by student t-test as shown (***Po0.01; **Po0.05).

In addition, ShYap knockdown in the Kios cells successfully inhibited the expression of H19 suggesting that Yap1 regulates the expression of H19 in ostesosarcoma (Figure 6c). Similarly, treatment with Hh inhibitor in primary tumor cells also inhibited H19 expression (Figure 6d). In contrast, forced expression of Yap1 into osteoblasts induced H19 expression, along with the activation of Hippo target genes (Figure 6e). Treatment with Shh or Hh agonist purmorphamine induced a subtle but significant H19 expression in wild-type primary osteoblasts as well (Figure 6f). Altogether, our data indicate that aberrant Hh signaling leads to overexpression of both Yap1 and H19, and that this is responsible for the pathogenesis and malignant transformation of osteoblastic osteosarcoma. DISCUSSION Here we show that upregulated Hh signaling in mature osteoblasts significantly induces osteoblastic osteosarcoma and that Yap1 and H19 are overexpressed and function downstream of Hh signaling as one of the underlying molecular mechanisms that leads to bone tumor formation. Furthermore, YAP1 overexpression is also observed in human patient specimens, suggesting that inhibition of Hh signaling and YAP1 expression are of significant therapeutic value. The initial bone overgrowth phenotypes observed in the homozygous Ptch1c/c; HOC-Cre mutant mice led us to hypothesize that Hh signaling is implicated in osteosarcoma development. Focal bone overgrowth tissues were observed in young mice (Supplementary Figure S1a–c), which is highly relevant to osteosarcoma in humans where the majority of patients are & 2013 Macmillan Publishers Limited

adolescents and youngsters.2 However, the mutant mice die within 3 months of birth, which makes it difficult to carry out detailed analysis of bone tumorigenesis. To circumvent this problem, we generated the Ptch1c/ þ ; p53 þ /  ;HOC-Cre mutant mice in which Hh signaling is partially upregulated in a p53 heterozygous background because previous studies showed that removal of p53 potentiates the onset of osteosarcoma in other mouse models such as retinoblastoma (Rb) mutant mice.3,4 Here, we used only p53 þ /  mice for our analysis because p53  /  mutant mice show a high frequency of spontaneous tumor development in other organs, which masks the specific effect of Hh signaling in bone tumor development. In our model, even with the heterozygous background of p53, accelerated onset of osteosarcoma development is observed. Some of the mutant mice also showed pulmonary metastasis at a late stage, which suggests that our mouse model is highly relevant to human osteosarcoma. In addition, tumor cells isolated from our mutant mice showed strong LacZ expression (Supplementary Figure S2a), further confirming that the tumors were derived from upregulated Hh signaling. We also analyzed osteoblasts isolated from the Ptch1c/c;HOC-Cre mutant mice alone, and we also found that Yap1 expression as well as Hippo downstream target genes were all highly upregulated without the influence from the manipulation of the p53 allele (Supplementary Figures S4e, g and h). Collectively, our data indicate that the initiation of osteosarcoma in our mouse model is mainly caused by upregulated Hh signaling. Another characteristic of our mouse model is that most of the developed bone tumors are mainly osteoblastic with massive osteoid formation, which is different from the preosteoblastrestricted deletion of p53 or Rb, where tumors are mainly Oncogene (2013) 1 – 10

Hh induced Yap1/H19 overexpression in osteosarcoma LH Chan et al

8 fibroblastic or of undifferentiated form.3,4 This clearly demonstrates the specific roles of Hh signaling in osteoblastic osteosarcoma development in which genetic manipulation in more mature osteoblasts (osteocalcin þ cells) instead of premature osteoblasts (osterix þ cells) or mesencymal stem cells (prx1 þ cells) may give rise to different subtypes of osteosarcoma. Although it is proposed that the origin of osteosarcoma is mainly derived from more primitive progenitor cells,36–38 a recent study using a transgenic mice that expressed the SV40 T antigen by osteocalcin promoter also presented solid osteosarcoma tumors with significant calcification.39 Further analysis revealed a deletion in Prkar1a that led to deregulated PKA signaling and Rankl overexpression. Indeed, we previously showed that Ptch1c/c;HOC-Cre mutant mice also display significant induction of Rankl expression.30 Many of the phenotypes from their transgenic mice were highly similar to our osteosarcoma mouse models in this study. Their findings together with ours indicate that osteoblastic osteosarcoma is originated from more mature osteoblasts. Previous studies showed that GLI2 is highly expressed in human osteosarcoma samples21,22 and treatment with Hh-specific inhibitor cyclopamine in human osteosarcoma cell lines inhibited cell proliferation.40 These findings hinted at the importance of aberrant Hh signaling in human osteosarcoma development. However, no genetically engineered animal model with detailed molecular mechanism is available. Here, we genetically demonstrated that Hh signaling induces osteosarcoma development through Yap1 and H19 overexpression. Our studies revealed that the Hippo pathway is a potential target for the treatment of osteosarcoma. Previous studies showed that 65% of neurofibromatosis 2 heterozygous mice developed osteosarcomalike disease41 and it has been suggested that neurofibromatosis 2/Merlin is one of the regulators acting upstream of Yap1.42 These findings are in agreement with our results suggesting that Hh signaling interacts with Hippo signaling for osteosarcoma development. In addition, a recent study showed that Amphiregulin, a secretory factor is also a downstream target of Hippo signaling in MCF10A breast cancer cells.34 Our data also confirmed that Amphiregulin is highly upregulated in both tumor tissues and primary osteosarcoma cells from our model (Supplementary Figure S4e and f). Altogether, our data support the notion that Hh signaling interacts with Hippo signaling for the pathogenesis of osteosarcoma. Although our study strongly suggests that Hh-induced Yap1 overexpression is associated with dysregulated Hippo signaling, it should be noted that Yap1 expression can be also regulated by other signaling pathways. It has been previously reported that Yap1 is a mediator of Src/Yes signaling and it functions to inhibit Runx2 activity during osteoblast differentiation.20 However, our study showed that Yap1 overexpression indeed promotes both Runx2 and Osteocalcin expression in our mouse model. These observations revealed the complexity of the regulation of Yap1 activity as well as its interaction with other key factors by different pathways. Runx2 is a key transcription factor for osteoblast differentiation, and it modulates the expression of cell cycle regulators in osteoblasts.43 The proliferation of Runx2  /  osteoblasts is much faster than that of the wild-type osteoblasts, suggesting that Runx2 has an antiproliferative function.44 However, enhanced expression of Runx2 is observed in many human osteosarcoma patient samples as well as human osteosarcoma cell lines,45–47 indicating that Runx2 may have an oncogenic function in tumor etiology. Other studies also showed that RUNX2 controls the expression of matrix metalloproteinases and VEGF that may be related to the process of cancer metastasis.48,49 A more detail investigation of the effect of other Hippo core components such as Mst1/2 on Runx2 activity may help to address this question. Our results demonstrate that H19 overexpression may contribute to the malignant transformation of osteosarcoma. H19 Oncogene (2013) 1 – 10

expression has been previously implicated in human osteosarcoma.29,50 Mutually exclusive loss of imprinting of H19 and IGF2 leads to reciprocal biallelic gene expression of each other and provides augmented growth-promoting signals to tumor growth. The regulation of H19 expression was shown to be regulated by epigenetics through methylation of the imprinting control region. The molecular regulation of H19 in tumorigenesis is still controversial, and its functional roles warrant further investigation. However, there is substantial evidence that H19 may function as an oncogene for the development of various cancers such as choriocarcinoma, hepatocellular carcinomas, esophageal cancer and ovarian cancers.51–54 Our studies provide the first evidence that H19 could be regulated by Hh signaling and Yap1 expression, and it may explain at least in part the pathogenesis of a subset of osteosarcomas. MATERIALS AND METHODS Mouse strains All animal experiments were performed according to procedures approved by the Animal Experimentation Ethics Committee of the Chinese University of Hong Kong. The Ptch1c/c;HOC-Cre30 and p53 þ /  31 mouse lines have been described previously. Both mutant mice were maintained in a mixed inbred C57BL/6 and 129/Sv background. Genotyping was done by PCR using genomic DNA isolated from the tails.

Histological analysis and western blot Primary antibodies used in immunohistochemistry include antiphosphohistone 3 (PH3) antibodies (Santa Cruz, Dallas, TX, USA) at 1:250 and Yap1 antibodies (Cell Signaling, Danvers, MA, USA) at 1:500. The signal was detected using Alexor Fluor conjugated secondary antibodies (Life Technologies, Grand Island, NY, USA) at 1:500 or biotinylated antibodies (Vector Labs, Burlingame, CA, USA) at 1:250 with DAB solution (SigmaAldrich, St Louis, MO, USA). b-Actin antibodies (Santa Cruz) at 1:5000 were used in western blot analysis for normalization.

Quantitative RT-PCR Total RNA was isolated using the TRIzol reagent (Life Technologies) according to the manufacturer’s instructions. Relative mRNA levels were examined using a Platinum SYBR Green kit (Life Technologies). Samples were normalized with b-tubulin or Gapdh expression.

Migration assay and transwell invasion assay Cells were pretreated with mitomycin C (4 mg/ml, Sigma-Aldrich) for 2 h prior to analysis. Cell migration was monitored for 12 h to 16 h by introduction of a scratch in confluent cells. For trans well invasion assay, BD Falcon Cell culture inserts were coated with BD Matrigel matrix (BD Bioscience, San Jose, CA, USA). Cells were seeded into each invasion chamber and incubated for 48 h. Non-invading cells were removed by cotton swabs and cells at the lower surface of the membrane were stained with crystal violet (Sigma-Aldrich).

Lentiviral transduction and luciferase assay ShYAP#3 and ShYAP#4 targeting different cDNA regions of Yap1 were constructed. Lentiviruses were prepared according to standard procedures. ShLuc or ShScramble were used as controls for comparison. Cells were incubated with individual lentivirus for 24 h and subsequent selection was performed with puromycin treatment for 4 days. Over 80% transduction efficiency was achieved. For luciferase assay, luciferase activity was measured 48 h after transfection according to the Dual-luciferase Reporter Assay System (Promega, Madison, WI, USA). The results are presented as the average±s.d. from three independent transfections.

Subcutaneous and intra-tibial injection 2  106 cells were mixed with 10% BD Matrigel in saline and injected subcutaneously to the rear flank of 6–8 weeks-old nude mouse using 27G needle (n ¼ 14). For intra-tibial injection, 5  105 cells were used and injected directly to the cavity of the proximal tibia from the knee joint of the hind limbs (n ¼ 18). Tumor growth was monitored for 1.5 months. & 2013 Macmillan Publishers Limited

Hh induced Yap1/H19 overexpression in osteosarcoma LH Chan et al

9 Tissue array Paraffin-embedded tissue array (SuperBioChips Lab, Seoul, Korea) with 54 human osteosarcoma patient samples from both sexes were examined for Yap1 expression by immunohistochemistry. Five normal bone tissues were included as controls. Standard procedures for immunohistochemistry were followed.

CONFLICT OF INTEREST The authors declare no conflict of interest.

ACKNOWLEDGEMENTS We thank Drs. Tin-lap Lee and David John Wilmshurst for critical comments on the manuscript. This work is supported by the Seed Fund of the School of Biomedical Sciences, The Chinese University of Hong Kong (4620504), Early Career Scheme of the Research Grant Council of the Hong Kong Special Administrative Region (CUHK479012), Direct Grant for research (CUHK2041745) and the Endowment Fund Research Grant of the United College, The Chinese University of Hong Kong (CA11189).

REFERENCES 1 Sweetnam R. Osteosarcoma. Brit J Hosp Med 1982; 28: 116–121. 2 Dorfman HD, Czerniak B. Bone cancers. Cancer 1995; 75: 203–210. 3 Berman SD, Calo E, Landman AS, Danielian PS, Miller ES, West JC et al. Metastatic osteosarcoma induced by inactivation of Rb and p53 in the osteoblast lineage. Proc Natl Acad SciUSA 2008; 105: 11851–11856. 4 Walkley CR, Qudsi R, Sankaran VG, Perry JA, Gostissa M, Roth SI et al. Conditional mouse osteosarcoma, dependent on p53 loss and potentiated by loss of Rb, mimics the human disease. Genes Dev 2008; 22: 1662–1676. 5 Calo E, Quintero-Estades JA, Danielian PS, Nedelcu S, Berman SD, Lees JA. Rb regulates fate choice and lineage commitment in vivo. Nature 2010; 466: 1110–1114. 6 Mutsaers AJ, Ng AJ, Baker EK, Russell MR, Chalk AM, Wall M et al. Modeling distinct osteosarcoma subtypes in vivo using Cre:lox and lineage-restricted transgenic shRNA. Bone 2013; 55: 166–178. 7 Kasper M, Regl G, Frischauf AM, Aberger F. GLI transcription factors: mediators of oncogenic Hedgehog signalling. Eur J Cancer 2006; 42: 437–445. 8 Rubin LL, de Sauvage FJ. Targeting the Hedgehog pathway in cancer. Nat Rev Drug Discov 2006; 5: 1026–1033. 9 Fults DW. Modeling medulloblastoma with genetically engineered mice. Neurosurg focus 2005; 19: E7. 10 Raffel C, Jenkins RB, Frederick L, Hebrink D, Alderete B, Fults DW et al. Sporadic medulloblastomas contain PTCH mutations. Cancer Res 1997; 57: 842–845. 11 Reifenberger J, Wolter M, Weber RG, Megahed M, Ruzicka T, Lichter P et al. Missense mutations in SMOH in sporadic basal cell carcinomas of the skin and primitive neuroectodermal tumors of the central nervous system. Cancer Res 1998; 58: 1798–1803. 12 Wetmore C. Sonic hedgehog in normal and neoplastic proliferation: insight gained from human tumors and animal models. Curr Opin Genet Dev 2003; 13: 34–42. 13 Fernandez LA, Northcott PA, Dalton J, Fraga C, Ellison D, Angers S et al. YAP1 is amplified and up-regulated in hedgehog-associated medulloblastomas and mediates Sonic hedgehog-driven neural precursor proliferation. Genes Dev 2009; 23: 2729–2741. 14 Lamar JM, Stern P, Liu H, Schindler JW, Jiang ZG, Hynes RO. The Hippo pathway target, YAP, promotes metastasis through its TEAD-interaction domain. Proc Natl Acad SciUSA 2012; 109: E2441–E2450. 15 Overholtzer M, Zhang J, Smolen GA, Muir B, Li W, Sgroi DC et al. Transforming properties of YAP, a candidate oncogene on the chromosome 11q22 amplicon. Proc Natl Acad SciUSA 2006; 103: 12405–12410. 16 Zhang X, George J, Deb S, Degoutin JL, Takano EA, Fox SB et al. The Hippo pathway transcriptional co-activator, YAP, is an ovarian cancer oncogene. Oncogene 2011; 30: 2810–2822. 17 Xu MZ, Yao TJ, Lee NP, Ng IO, Chan YT, Zender L et al. Yes-associated protein is an independent prognostic marker in hepatocellular carcinoma. Cancer 2009; 115: 4576–4585. 18 Muramatsu T, Imoto I, Matsui T, Kozaki K, Haruki S, Sudol M et al. YAP is a candidate oncogene for esophageal squamous cell carcinoma. Carcinogenesis 2011; 32: 389–398. 19 Yagi R, Chen LF, Shigesada K, Murakami Y, Ito YA. WW domain-containing yes-associated protein (YAP) is a novel transcriptional co-activator. EMBO J 1999; 18: 2551–2562.

& 2013 Macmillan Publishers Limited

20 Zaidi SK, Sullivan AJ, Medina R, Ito Y, van Wijnen AJ, Stein JL et al. Tyrosine phosphorylation controls Runx2-mediated subnuclear targeting of YAP to repress transcription. EMBO J 2004; 23: 790–799. 21 Hirotsu M, Setoguchi T, Sasaki H, Matsunoshita Y, Gao H, Nagao H et al. Smoothened as a new therapeutic target for human osteosarcoma. Molecular cancer 2010; 9: 5. 22 Nagao H, Ijiri K, Hirotsu M, Ishidou Y, Yamamoto T, Nagano S et al. Role of GLI2 in the growth of human osteosarcoma. J Pathol 2011; 224: 169–179. 23 Matouk IJ, Abbasi I, Hochberg A, Galun E, Dweik H, Akkawi M. Highly upregulated in liver cancer noncoding RNA is overexpressed in hepatic colorectal metastasis. Eur JGastroen Hepat 2009; 21: 688–692. 24 Ponting CP, Oliver PL, Reik W. Evolution and functions of long noncoding RNAs. Cell 2009; 136: 629–641. 25 Rinn JL, Kertesz M, Wang JK, Squazzo SL, Xu X, Brugmann SA et al. Functional demarcation of active and silent chromatin domains in human HOX loci by noncoding RNAs. Cell 2007; 129: 1311–1323. 26 Schmidt LH, Spieker T, Koschmieder S, Schaffers S, Humberg J, Jungen D et al. The long noncoding MALAT-1 RNA indicates a poor prognosis in non-small cell lung cancer and induces migration and tumor growth. J Thorac Oncol 2011; 6: 1984–1992. 27 Wang J, Liu X, Wu H, Ni P, Gu Z, Qiao Y et al. CREB up-regulates long non-coding RNA, HULC expression through interaction with microRNA-372 in liver cancer. Nucleic Acids Res 2010; 38: 5366–5383. 28 Pachnis V, Brannan CI, Tilghman SM. The structure and expression of a novel gene activated in early mouse embryogenesis. EMBO J 1988; 7: 673–681. 29 Ulaner GA, Vu TH, Li T, Hu JF, Yao XM, Yang Y et al. Loss of imprinting of IGF2 and H19 in osteosarcoma is accompanied by reciprocal methylation changes of a CTCF-binding site. Hum Mol Gen 2003; 12: 535–549. 30 Mak KK, Bi Y, Wan C, Chuang PT, Clemens T, Young M et al. Hedgehog signaling in mature osteoblasts regulates bone formation and resorption by controlling PTHrP and RANKL expression. Dev Cell 2008; 14: 674–688. 31 Venkatachalam S, Tyner SD, Pickering CR, Boley S, Recio L, French JE et al. Is p53 haploinsufficient for tumor suppression? Implications for the p53 þ /  mouse model in carcinogenicity testing. Toxicol Pathol 2001; 29: 147–154. 32 Mak KK, Chen MH, Day TF, Chuang PT, Yang Y. Wnt/beta-catenin signaling interacts differentially with Ihh signaling in controlling endochondral bone and synovial joint formation. Development 2006; 133: 3695–3707. 33 Zhang L, Ren F, Zhang Q, Chen Y, Wang B, Jiang J. The TEAD/TEF family of transcription factor Scalloped mediates Hippo signaling in organ size control. Dev Cell 2008; 14: 377–387. 34 Zhang J, Ji JY, Yu M, Overholtzer M, Smolen GA, Wang R et al. YAP-dependent induction of amphiregulin identifies a non-cell-autonomous component of the Hippo pathway. Nat Cell Biol 2009; 11: 1444–1450. 35 Jones KB, Salah Z, Del Mare S, Galasso M, Gaudio E, Nuovo GJ et al. miRNA signatures associate with pathogenesis and progression of osteosarcoma. Cancer Res 2012; 72: 1865–1877. 36 Li N, Yang R, Zhang W, Dorfman H, Rao P, Gorlick R. Genetically transforming human mesenchymal stem cells to sarcomas: changes in cellular phenotype and multilineage differentiation potential. Cancer 2009; 115: 4795–4806. 37 Mohseny AB, Szuhai K, Romeo S, Buddingh EP, Briaire-de Bruijn I, de Jong D et al. Osteosarcoma originates from mesenchymal stem cells in consequence of aneuploidization and genomic loss of Cdkn2. J Pathol 2009; 219: 294–305. 38 Tang N, Song WX, Luo J, Haydon RC, He TC. Osteosarcoma development and stem cell differentiation. Clin Orthop Relat Res 2008; 466: 2114–2130. 39 Molyneux SD, Di Grappa MA, Beristain AG, McKee TD, Wai DH, Paderova J et al. Prkar1a is an osteosarcoma tumor suppressor that defines a molecular subclass in mice. J Clin Invest 2010; 120: 3310–3325. 40 Warzecha J, Gottig S, Chow KU, Bruning C, Percic D, Boehrer S et al. Inhibition of osteosarcoma cell proliferation by the Hedgehog-inhibitor cyclopamine. J Chemotherapy 2007; 19: 554–561. 41 McClatchey AI, Saotome I, Mercer K, Crowley D, Gusella JF, Bronson RT et al. Mice heterozygous for a mutation at the Nf2 tumor suppressor locus develop a range of highly metastatic tumors. Genes Dev 1998; 12: 1121–1133. 42 Zhang N, Bai H, David KK, Dong J, Zheng Y, Cai J et al. The Merlin/NF2 tumor suppressor functions through the YAP oncoprotein to regulate tissue homeostasis in mammals. Dev Cell 2010; 19: 27–38. 43 Galindo M, Pratap J, Young DW, Hovhannisyan H, Im HJ, Choi JY et al. The bonespecific expression of Runx2 oscillates during the cell cycle to support a G1related antiproliferative function in osteoblasts. J Biol Chem 2005; 280: 20274–20285. 44 Teplyuk NM, Galindo M, Teplyuk VI, Pratap J, Young DW, Lapointe D et al. Runx2 regulates G protein-coupled signaling pathways to control growth of osteoblast progenitors. J Biol Chem 2008; 283: 27585–27597.

Oncogene (2013) 1 – 10

Hh induced Yap1/H19 overexpression in osteosarcoma LH Chan et al

10 45 Lu XY, Lu Y, Zhao YJ, Jaeweon K, Kang J, Xiao-Nan L et al. Cell cycle regulator gene CDC5L, a potential target for 6p12-p21 amplicon in osteosarcoma. Molecular cancer research: MCR 2008; 6: 937–946. 46 Sadikovic B, Thorner P, Chilton-Macneill S, Martin JW, Cervigne NK, Squire J et al. Expression analysis of genes associated with human osteosarcoma tumors shows correlation of RUNX2 overexpression with poor response to chemotherapy. BMC cancer 2010; 10: 202. 47 Andela VB, Siddiqui F, Groman A, Rosier RN. An immunohistochemical analysis to evaluate an inverse correlation between Runx2/Cbfa1 and NF kappa B in human osteosarcoma. J Clin Pathol 2005; 58: 328–330. 48 Pratap J, Javed A, Languino LR, van Wijnen AJ, Stein JL, Stein GS et al. The Runx2 osteogenic transcription factor regulates matrix metalloproteinase 9 in bone metastatic cancer cells and controls cell invasion. Mol Cell Biol 2005; 25: 8581–8591.

49 Won KY, Park HR, Park YK. Prognostic implication of immunohistochemical Runx2 expression in osteosarcoma. Tumori 2009; 95: 311–316. 50 Ulaner GA, Yang Y, Hu JF, Li T, Vu TH, Hoffman AR. CTCF binding at the insulin-like growth factor-II (IGF2)/H19 imprinting control region is insufficient to regulate IGF2/H19 expression in human tissues. Endocrinology 2003; 144: 4420–4426. 51 Arima T, Matsuda T, Takagi N, Wake N. Association of IGF2 and H19 imprinting with choriocarcinoma development. Cancer Genet Cytogen 1997; 93: 39–47. 52 Matouk IJ, DeGroot N, Mezan S, Ayesh S, Abu-lail R, Hochberg A et al. The H19 non-coding RNA is essential for human tumor growth. PloS one 2007; 2: e845. 53 Hibi K, Nakamura H, Hirai A, Fujikake Y, Kasai Y, Akiyama S et al. Loss of H19 imprinting in esophageal cancer. Cancer Res 1996; 56: 480–482. 54 Tanos V, Ariel I, Prus D, De-Groot N, Hochberg A. H19 and IGF2 gene expression in human normal, hyperplastic, and malignant endometrium. Int J Gynecol 2004; 14: 521–525.

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

Oncogene (2013) 1 – 10

& 2013 Macmillan Publishers Limited