SEARCH

SEARCH BY CITATION

Keywords:

  • osteosarcoma;
  • hedgehog signaling;
  • ligand-dependent;
  • IPI-926;
  • microenvironment

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

BACKGROUND

During development, the Hedgehog pathway plays important roles regulating the proliferation and differentiation of chondrocytes, providing a template for growing bone. In this study, the authors investigated the components of dysregulated Hedgehog signaling as potential therapeutic targets for osteosarcoma.

METHODS

Small-molecule agonists and antagonists that modulate the Hedgehog pathway at different levels were used to investigate the mechanisms of dysregulation and the efficacy of Hedgehog blockade in osteosarcoma cell lines. The inhibitory effect of a small-molecule Smoothened (SMO) antagonist, IPI-926 (saridegib), also was examined in patient-derived xenograft models.

RESULTS

An inverse correlation was identified in osteosarcoma cell lines between endogenous glioma-associated oncogene 2 (GLI2) levels and Hedgehog pathway induction levels. Cells with high levels of GLI2 were sensitive to GLI inhibition, but not SMO inhibition, suggesting that GLI2 overexpression may be a mechanism of ligand-independent activation. In contrast, cells that expressed high levels of the Hedgehog ligand gene Indian hedgehog (IHH) and the target genes patched 1 (PTCH1) and GLI1 were sensitive to modulation of both SMO and GLI, suggesting ligand-dependent activation. In 2 xenograft models, active autocrine and paracrine, ligand-dependent Hedgehog signaling was identified. IPI-926 inhibited the Hedgehog signaling interactions between the tumor and the stroma and demonstrated antitumor efficacy in 1 of 2 ligand-dependent models.

CONCLUSIONS

The current results indicate that both ligand-dependent and ligand-independent Hedgehog dysregulation may be involved in osteosarcoma. It is the first report to demonstrate Hedgehog signaling crosstalk between the tumor and the stroma in osteosarcoma. The inhibitory effect of IPI-926 warrants additional research and raises the possibility of using Hedgehog pathway inhibitors as targeted therapeutics to improve treatment for osteosarcoma. Cancer 2014;120:537–547. © 2013 American Cancer Society.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

Osteosarcoma is the most common primary bone cancer in children and adolescents.[1] Most osteosarcomas are high-grade tumors and are associated with a poor prognosis. For the 10% to 20% of patients who have detectable metastases to the lungs at diagnosis, few will be cured despite aggressive multimodal therapy.[2] The combination of chemotherapy and surgery has significantly improved the long-term survival of patients with osteosarcoma who present with localized disease up to 50% or 60%.[2, 3] However, survival has plateaued over the past 20 years because of a lack of treatment improvements. No targeted therapy for osteosarcoma currently exists; hence, the objective of this study was to gain a better understanding of its underlying molecular alterations to help identify superior therapeutic interventions.

The Hedgehog (Hh) pathway regulates the proliferation and differentiation of tissues and organs during development. Components of the Hh pathway are tightly regulated to control signaling.[4, 5] Among the Sonic (SHH), Indian (IHH), and Desert (DHH) Hh ligands in the vertebrate system, the IHH ligands are the main ligands for endochondral ossification during long bone development.[6, 7] The Hh ligand signals through the patched (PTCH) receptor to relieve its inhibition of the smoothened (SMO) receptor, which, through a series of intracellular signaling cascades, results in the activation of glioma-associated oncogene (GLI) transcription regulators.[4] Aberrant activation of the Hh pathway, through ligand-dependent or ligand-independent mechanisms, has been reported in numerous cancers.[5] Several small-molecule antagonists of the Hh pathway are being explored clinically. The approval of an SMO antagonist, vismodegib, for the treatment of locally advanced and metastatic basal cell carcinomas brings promise for other cancers that have dysregulated Hh signaling.[8]

We previously identified high levels of coexpression of IHH and the Hh-regulated targets PTCH1 and GLI1 in a subset of osteosarcoma samples (Lo et al, unpublished results). For the current study, we investigated the mechanisms of Hh pathway dysregulation in osteosarcoma and the potential of Hh-inhibitory therapy for treatment. Dysregulated Hh signaling was examined in osteosarcoma cell lines and patient-derived xenograft models using small-molecule modulators of the Hh pathway.[9-12] The efficacy of a small-molecule inhibitor of SMO, IPI-926 (saridegib),[13-17] was studied in xenograft models, which also provided the opportunity to characterize Hh signaling interactions between the tumor and its microenvironment, as demonstrated in other cancers.[13, 18-20]

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

Cell Culture

Osteosarcoma cell lines were purchased from the American Type Culture Collection (Manassas, Va) and were cultured as recommended and authenticated by short tandem repeat profiling. Mouse embryonic mesoderm fibroblasts, C3H10T1/2 cells, and Ptch−/− mouse embryonic fibroblast cells were maintained in Dulbecco modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS).

Hedgehog Pathway Agonist and Antagonist Treatments

The following molecules were used for treatment: 2 μM purmorphamine (Calbiochem-Novabiochem Corporation, San Diego, Calif) in dimethyl sulfoxide (DMSO), 10 nM synthetic SMO agonist (SAG) (Alexis Biochemicals, San Diego, Calif) in DMSO, 5 μM cyclopamine (BioMol, Plymouth Meeting, Pa) in 100% ethanol (EtOH), and 5 μM GANT61 (Developmental Therapeutics Program, National Cancer Institute/National Institutes of Health, Bethesda, Md) in DMSO. Cells were treated for 48 hours or 72 hours under serum-starved conditions (0.5% FBS). Treatments were performed in duplicate and were thrice repeated. To test the efficacy of IPI-926 in inhibiting Hh signaling, MG-63 cells were treated with 10% Shh-conditioned media (CM) either alone or in combination with 500 nM IPI-926 in 0.5% FBS media after 5 days of serum starvation. Shh-CM:293 EcR Shh (JHU-64) cells were cultured in complete DMEM medium with 400 μg/mL G418 until they reached 80% confluence. The medium was replaced with DMEM containing 2% FBS for 24 hours, at which point the CM was collected and filtered through a 0.22-μm filter.

Small-Interference Transfection

GLI1 expression was silenced in Saos-2 cells with GLI1 small-interference RNAs (siRNAs) (siRNA 115641 and siRNA 115642; Ambion Inc., Austin, Tex) using the reverse transfection method with siPORT Amine (Ambion Inc.). Cell lysates were collected 48 hour post-transfection for protein quantification and 72 hour post-transfection for cell proliferation assays. Treatments were performed in duplicate and were thrice repeated.

Cell Proliferation Assays

Cell proliferation assays were carried out using a dimethyl thiazolyl diphenyl tetrazolium salt (3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide [MTT]) colorimetric cell proliferation kit (Roche Diagnostics Corporation, Indianapolis, Ind) or were fixed with 0.4% paraformaldehyde, stained with 4′,6-diamidino-2-phenylindole, and analyzed with the In Cell Analyzer 2000 (GE Healthcare, Rahway, NJ). Experiments were performed in triplicate and were thrice repeated.

Western Blot Analysis

Membranes were blotted using GLI1 antibodies (ab49314; Abcam plc, Cambridge, United Kingdom), GLI2 antibodies (ab7195; Abcam plc), PTCH1 antibodies (sc-6149; Santa Cruz Biotechnology, Inc., Santa Cruz, Calif), and β-actin (Sigma Chemical Company, St. Louis, Mo) as a loading control, followed by secondary conjugates (horseradish peroxidase-donkey antibodies; Jackson Immunochemicals, West Grove, Pa).

Patient-Derived Osteosarcoma Xenograft Models and IPI-926 Treatment

This research was performed with the approval of the appropriate human and animal ethics committees at our institutions. Patients provided a signed consent form before study entry. Fresh samples that were collected during open surgical biopsies were divided into equal sizes (0.2 × 0.2 cm) and were bilaterally, subcutaneously implanted into the flanks of immunodeficient nonobese diabetic/severe combined immunodeficiency (NOD/SCID) mice to permit additional tumor growth for 6 to 8 weeks. Four independent xenograft models (models A, B, C, and D) were generated from the primary tumors of 4 patients who had various clinical and pathologic characteristics (Table 1). Tissue samples from the site of tumor implantation were collected from 4 nonengrafted mice to determine basal levels of Hh signaling.

Table 1. Clinical and Pathologic Characteristics Associated With Patient-Derived Osteosarcoma Xenograft Models
CharacteristicXenograft Model AXenograft Model BXenograft Model CXenograft Model Da
  1. Abbreviations: ANED, alive with no evidence of disease; DOD, died of disease.

  2. a

    Xenograft model D was included because the tumor was resistant to conventional chemotherapy.

Systemic disease statusMetastasis at diagnosisLocalized disease at diagnosisMetastasis at diagnosisLocalized disease at diagnosis
Tumor gradeIIIIIIIIII
Tumor siteProximal femurMid-forearmDistal femurProximal tibia
Age at diagnosis, y169929
SexMaleFemaleFemaleMale
Patient statusUnknownANEDDODANED
SubtypeChondroblasticOsteoblasticOsteoblasticFibroblastic central

Once tumors were grossly visible in the mice (after 4-6 weeks), 5 mice received 40 mg/kg of IPI-926 orally, and another 5 mice received the vehicle control. Xenograft A mice were treated 5 times a week. Eighteen treatments were administered by the endpoint, at which time the tumor burden became overbearing for the mice. Because of the aggressive nature of xenograft A, the subsequent xenografts received treatment 7 times a week. Twenty treatments were administered to xenograft B mice, 37 treatments were administered to xenograft C mice, and 31 treatments were administered to xenograft D mice. An external caliper was used to determine tumor size weekly in mouse xenografts B, C, and D.

Quantitative Real-Time Polymerase Chain Reaction Analysis

Samples were snap-frozen immediately after harvesting. Total RNA was extracted using TRIzol (Invitrogen, Carlsbad, Calif). Complementary DNA was generated from total RNA as previously described.[21] Quantitative real-time polymerase chain reaction analysis was performed on an ABI Prism 7000 Sequence Detector using SYBR Green Master Mix (Applied Biosystems, Foster City, Calif). The absolute standard curve method was used for quantification. Signal-transducing adaptor molecule 2 (STAM2), glyceraldehyde-3-phosphate dehydrogenase (GAPDH), and asparagine synthetase (ASNS) were used as housekeeping genes. Transcript levels were normalized to the levels of the housekeeping gene.

Histology, Immunohistochemistry, and Immunofluorescence Analyses

Tissue was stained using hematoxylin and eosin for routine morphologic analysis. Proliferation was assessed using immunohistochemistry for Ki-67 (MIB-1; Dako, Carpenteria, Calif) at 1:200 dilution overnight at 4°C. The 3,3′-diaminobenzidine peroxidase substrate kit (Vector Laboratories, Burlingame, Calif) was used for visualization. The Ki-67 proliferation index was assessed blindly and was estimated based on the percentage of positive tumor cells in the tumor section. Apoptotic cells were evaluated by terminal deoxynucleotidyl transferase dUTP nick-end labeling (TUNEL) (EMD Millipore, Billerica, Mass). Each section was assessed blindly, and the total number of positive cells in 10 high-power fields of view (×400 magnification; field diameter, 0.55 mm) was counted. A sensitive human GLI1 rabbit monoclonal antibody (Cell Signaling Technology, Inc., Beverly, Mass) was used to observe nuclear GLI1 in tissue sections (1:200 dilution overnight at room temperature). Heat-induced epitope retrieval was carried out. A TSA-Cy5 kit (Perkin Elmer, Waltham, Mass) was used to amplify and observe the GLI1 signal. Images were captured at ×40 magnification using the TissueFAXS system (TissueGnostics, Tarzana, Calif).

Statistical Analyses

All data are represented as mean ± standard deviation values. Statistical calculations were performed in Microsoft Excel (Microsoft Corporation, Redmond, Wash). Differences between groups were analyzed using Student t tests. Pearson correlation coefficients were calculated using Matlab (The MathWorks, Inc., Natick, Mass).

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

Inverse Correlation Between Endogenous GLI2 mRNA Levels and Hedgehog Pathway Induction Levels in Osteosarcoma Cell Lines

Variable levels of IHH, PTCH1, GLI1, GLI2, and SMO were observed in osteosarcoma cell lines (Fig. 1A-E). A positive correlation was observed between the levels of Indian Hedgehog ligand, IHH, and the target genes (PTCH1 and GLI1) as well as the Smoothened receptor, SMO, with Saos-2 cells exhibiting the highest levels of expression. Although both GLI1 and GLI2 function as transcriptional activators of the Hh pathway, their expression levels were not the same in all the cell lines (Fig. 1C,D). Saos-2 cells expressed high levels of GLI1 but low levels of GLI2, whereas U2OS and HOS cells expressed high levels of GLI2 but low levels of GLI1. Western blot analysis revealed that the relative protein expression of PTCH1 (Fig. 1F), GLI1 (Fig. 1G), and GLI2 (Fig. 1H) was similar to the messenger RNA levels in the cell lines.

image

Figure 1. Expression levels of Hedgehog pathway genes are illustrated in 6 osteosarcoma cell lines. Transcript levels of (A) Indian hedgehog (IHH), (B) patched 1 (PTCH1), (C) glioma-associated oncogene 1 GLI1, (D) GLI2, and (E) smoothened (SMO) as well as protein levels of (F) PTCH1, (G) GLI1, and (H) GLI2 are illustrated in osteosarcoma cell lines.

Download figure to PowerPoint

GLI1 transcript levels are being used routinely to determine Hh activities and for pharmacokinetic and pharmacodynamic analyses of Hh modulators. GLI1 amplifies Hh signaling at the transcript level, which makes GLI1 messenger RNA levels the most reliable indicators of pathway activity.[22-24] In this study, GLI1 levels were measured to examine the degree of Hh pathway activation through the agonists purmorphamine and SAG in osteosarcoma cell lines. Upon treatment with 2 μM purmorphamine, Saos-2, KHOS, MNNG, and MG-63 cells exhibited induction of GLI1 levels, and MG-63 cells had the highest induction levels (from 3.5-fold to 4-fold) (Fig. 2A). Similar results were observed at the protein level (Fig. 2B) and when SAG was used (data not shown). It is noteworthy that the cells with high endogenous levels of GLI2, such as U2OS and HOS cells, were insensitive to pathway induction. A strong inverse correlation (r = −0.89; P = .02) was observed between the levels of GLI1 induction and endogenous GLI2 (Fig. 2C). These results suggest that U2OS and HOS cells may have constitutive Hh signaling caused by high GLI2 levels and, thus, are insensitive to Hh agonists that act upstream at the receptor level.

image

Figure 2. Activation of the Hedgehog pathway by purmorphamine is illustrated in osteosarcoma cell lines. (A) The fold induction of glioma-associated oncogene 1 (GLI1) transcript levels in osteosarcoma cell lines by 2 μM purmorphamine is illustrated relative to the transcript levels in their control counterparts. (B) GLI1 protein expression of cell lines after purmorphamine treatment is illustrated. DMSO indicates dimethyl sulfoxide. (C) This chart illustrates the inverse correlation between endogenous GLI2 expression and GLI1 induction levels in osteosarcoma (OS) cell lines.

Download figure to PowerPoint

Inhibition of Activated Hedgehog Signaling in Osteosarcoma Cell Lines

To determine whether aberrant Hh signaling can be inhibited in osteosarcoma, an SMO antagonist, cyclopamine,[9] and a GLI antagonist, GANT61,[11] were used to block Hh signaling in the osteosarcoma cell lines (Saos-2, HOS, and U2OS) that displayed high levels of the GLI activators GLI1 and/or GLI2. Decreased GLI1 transcript and protein levels were observed only in 5 μM cyclopamine-treated Saos-2 cells but not in U2OS or HOS cells (Fig. 3A). A corresponding decrease in proliferation was observed in treated Saos-2 cells but not in U2OS or HOS cells (Fig. 3B). Treatment with 5 μM GANT61 resulted in decreased GLI1 levels in Saos-2 and HOS cells but not in U2OS cells (Fig. 3C). A corresponding decrease in cell proliferation was observed in GANT61-treated Saos-2 and HOS cells (Fig. 3D). To validate the results produced with GANT61, siRNAs were used to knockdown GLI1 expression in Saos-2 cells. GLI1 siRNAs significantly decreased the endogenous GLI1 protein expression in Saos-2 cells (Fig. 3E). Decreased cell proliferation was also observed with GLI1 siRNA1 (Fig. 3F).

image

Figure 3. Inhibition of activated Hedgehog (Hh) signaling is illustrated in osteosarcoma cell lines. (A) Cyclopamine (Cyclo) (5 μM) decreased glioma-associated oncogene 1 (GLI1) expression in Saos-2 cells, but not in U2OS or HOS cells, at both (Top) the transcript level and (Bottom) the protein level. EtOH indicates ethanol. (B) Decreased cell proliferation is illustrated in Saos-2 cells after cyclopamine treatment. (C) The GLI antagonist GANT61 (5 μM) decreased GLI1 levels in Saos-2 and HOS cells, but not in U2OS cells, at both (Top) the transcript level and (Bottom) the protein level. DMSO indicates dimethyl sulfoxide. (D) Decreased cell proliferation is illustrated in Saos-2 and HOS cells after GANT61 treatment. (E) GLI1 small-interference RNAs (siRNAs) decreased the protein levels of GLI1 in Saos-2 cells. Ctrl indicates control. (F) Decreased cell growth is illustrated in GLI1-silenced Saos-2 cells with siRNA1. A single asterisk indicates P < .05; double asterisks, P = .01 (n = 3). (G) MG-63 cells treated with Sonic hedgehog (Shh)-conditioned media (CM) resulted in increased GLI1 levels, which were inhibited by IPI-926 (saridegib) treatment.

Download figure to PowerPoint

Saos-2 cells expressed high levels of IHH and the target genes PTCH1 and GLI1. The finding that Hh signaling could be effectively activated at the receptor level and inhibited at the receptor and transcriptional levels provided further support that the dysregulation in Saos-2 cells is likely ligand-dependent. In contrast, HOS cells exhibited high levels of GLI2 but low levels of IHH and GLI1 and were sensitive only to inhibition at the transcript level but not at the receptor level. These data support the purmorphamine treatment data and suggest that Hh dysregulation in HOS cells is likely ligand-independent, resulting from high levels of GLI2.

Although it has been demonstrated that cyclopamine inhibits Hh signaling and tumor growth of several cancers, its nonspecificity and poor pharmacokinetic properties have limited further clinical development of cyclopamine as a therapeutic agent.[25] High concentrations (≥10 μM) of cyclopamine also caused non-Hh–specific toxicity in HOS and U2OS cells (data not shown). Therefore, we examined a semisynthetic analog of cyclopamine, IPI-926, with improved structure and superior potency and pharmacokinetic properties.[14] To test the efficacy of IPI-926 in inhibiting activated Hh signaling, MG-63 cells were treated with Shh-CM alone and in combination with IPI-926. Activated Hh signaling, as observed by increased GLI1 levels, was identified with Shh-CM treatment, and this activation was abolished by IPI-926 (Fig. 3G).

Active Ligand-Dependent Autocrine and Paracrine Hedgehog Signaling in Osteosarcoma Xenograft Models

Patient-derived osteosarcoma xenograft models allow for the investigation of Hh pathway signaling interactions between the human tumor compartments and the mouse stromal compartments, which consist of endothelial cells, pericytes, and fibroblasts and presumably could support tumor growth. Although it would be preferable to use models that demonstrate high levels of Hh activity, because of limited tumor availability and xenograft models, 4 independent models (xenografts A, B, C, and D) with differential levels of Hh activities and clinical characteristics were used in this study (Table 1).

By using human-specific primers, variable levels of IHH, PTCH1, and GLI1 were observed in the tumors from the xenograft models (Fig. 4A). Coexpression of IHH and the target genes PTCH1 and GLI1 was observed in xenografts A and C, with xenograft A exhibiting the highest levels of expression, suggesting autocrine ligand-dependent activation.

image

Figure 4. Hedgehog pathway expression is illustrated in human tumor and mouse stroma from 4 independent osteosarcoma xenograft models (models A, B, C, and D). (A) Human-specific primers were used to quantify expression in human tumors. (B) Mouse-specific primers were used to quantify expression in mouse stroma. Expression in the stroma was normalized to the expression observed in muscles from nonengrafted nonobese diabetic/severe combined immunodeficiency mice. Downward arrows represent minute values that cannot be observed in the graph.

Download figure to PowerPoint

Mouse-specific primers were used to identify Hh activity in the tumor-associated mouse stroma (Fig. 4B). It is noteworthy that the expression of IHH was identified exclusively in the human tumor, whereas the expression of Shh was identified exclusively in the mouse stroma. High levels of Shh, Ptch1, and Gli1 also were identified in the mouse stroma of xenografts A and C. The coexpression of IHH/Shh and the targets PTCH1/Ptch1 and GLI1/Gli1 in both the tumor and the stroma of xenografts A and C suggests autocrine ligand-dependent activation in both compartments. Because the basal levels of Shh are low in nonengrafted mice, the high levels of Shh in the mouse stroma suggest that the elevated expression is likely caused by paracrine Hh signaling. In contrast, xenograft B exhibited low levels of all genes, and xenograft D exhibited low levels of the positive regulators IHH and GLI1 in the tumor, suggesting the absence of ligand-dependent signaling.

Specific Inhibition of the Ligand-Dependent Hedgehog Pathway by IPI-926 in Osteosarcoma Xenograft Models

We used IPI-926, a clinical candidate, to evaluate Hh pathway inhibition as a possible therapeutic intervention for osteosarcoma in our patient-derived xenograft models and observed that IPI-926 was effective at inhibiting ligand-dependent Hh pathway signaling in xenografts A and C. IPI-926-treated xenografts A and C exhibited decreased levels of GLI1 and PTCH1 in both the tumor and the stroma (Fig. 5A). By using a human-specific GLI1 antibody, decreased GLI1 protein expression also was observed in the tumors of IPI-926-treated xenografts A and C (Fig. 5B). In contrast, xenografts B and D, which did not exhibit active ligand-dependent Hh signaling, were unresponsive to IPI-926 treatment (data not shown).

image

Figure 5. Specific inhibition of ligand-dependent Hedgehog (Hh) activity by IPI-926 (saridegib) was investigated in xenograft tumors and stroma. (A) Xenografts A and C, which demonstrated ligand-dependent Hh signaling, were sensitive to IPI-926 treatment, as demonstrated by the decreased levels of glioma-associated oncogene 1 (GLI1) and patched 1 (PTCH1) in tumor and stroma. A single asterisk indicates P ≤ .003; double asterisks, P = .05 (n = 3). (B) Micrographs reveal staining by immunofluorescence for human GLI1 (green) with 4′,6-diamidino-2-phenylindole dihydrochloride-stained nuclei (blue) in control (Ctrl)-treated versus IPI-926–treated tumors from xenografts A and C.

Download figure to PowerPoint

Inhibition of Hedgehog Signaling by IPI-926 and Antitumor Efficacy

Inhibition of Hh signaling by IPI-926 treatment resulted in significantly decreased tumor weight and volume of xenograft C (Table 2). Weekly monitoring of tumor size revealed that this was likely because of decreased tumor growth (data not shown). A trend toward decreased tumor weight was observed in treated xenograft A. The insensitivity of xenografts B and D to IPI-926 resulted in no tumor volume or weight change.

Table 2. Weight and Volume of IPI-926–Treated Tumors Versus Control-Treated Tumors From Xenograft Models at the Treatment Endpoint
 Xenograft Model AXenograft Model BXenograft Model CaXenograft Model D
VariableCtrlIPI-926PCtrlIPI-926PCtrlIPI-926PCtrlIPIP
  1. Abbreviations: Ctrl, control; NA, not applicable.

  2. a

    Boldface P values indicate statistical significance.

Weight, g2.491.8.234.093.76.363.342.05.050.540.41.22
Volume, cm3NANA6.616.4.435.192.95.040.590.37.13

Proliferation and Apoptosis Effects of Hedgehog Signaling Inhibition by IPI-926 in Osteosarcoma Xenograft Models

Each of the 4 osteosarcoma xenograft models had distinguishing morphologic attributes that resembled the parent primary tumor. Hematoxylin and eosin-stained sections of the tumors were reviewed by a pathologist with expertise in bone pathology (B.C.D.) who was blinded to treatment to determine whether the inhibition of Hh signaling resulted in distinct morphologic changes. No appreciable histologic differences were identified between IPI-926-treated tumors and control-treated tumors in terms of nuclear pleomorphism, mitotic activity, necrosis, or extracellular matrix production (data not shown).

The Hh pathway is important for cellular processes, such as apoptosis, proliferation, and differentiation.[6, 26-28] Immunohistochemical staining of Ki-67 and TUNEL were used to assess the effect on proliferation and apoptosis, respectively, by IPI-926. Semiquantitative analysis suggested increased apoptosis in IPI-926-treated tumors of xenograft C (Fig. 6A). A trend toward decreased proliferation was observed in IPI-926-treated xenografts A, B, and C (Fig. 6B).

image

Figure 6. The effects of apoptosis and proliferation in xenograft models upon Hedgehog signaling inhibition by IPI-926 (saridegib) treatment are illustrated. (A) Top: The dot plot illustrates the distribution of terminal deoxynucleotidyl transferase dUTP nick-end labeling (TUNEL)-positive cells in IPI-926–treated and control-treated tumors to evaluate apoptosis. Xenograft C (upward triangles) exhibited a trend toward increased apoptosis in IPI-926–treated tumors. Bottom: Representative images of cells indicate low versus high levels of TUNEL staining in xenograft C tumors (upward triangles). (B) Top: The dot plot illustrates distribution according to the percentage of Ki-67–positive cells in IPI-926–treated and control-treated tumors to evaluate proliferation. A trend toward decreased proliferation was observed in xenograft models A (circles), B (squares), and C (upward triangles). Bottom: Representative images of cells indicate relative low versus high levels of Ki-67 staining in xenograft C tumors. Downward triangles indicate xenograft model D.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

The purpose of this study was to characterize dysregulated Hh signaling and evaluate the inhibition of constitutively activated Hh signaling as a novel therapeutic target in osteosarcoma. Using small-molecule agonists and antagonists that target different components of the Hh pathway, we observed not only that there are different mechanisms of Hh dysregulation in osteosarcoma but that the microenvironment also plays a significant role in promoting aberrant Hh signaling. Although the involvement of Hh signaling has been implicated previously in osteosarcoma,[29-33] to our knowledge, this is the first report demonstrating Hh signaling interactions between the tumor and its microenvironment.

Saos-2 cells, which exhibited high levels of IHH, GLI1, and PTCH1, were sensitive to the modulation of both SMO and GLI, suggesting that Saos-2 cells have autocrine ligand-dependent Hh pathway activation. It was also interesting to note that cells (eg, MG-63) with low levels of Hh activities were sensitive to SMO agonists, suggesting that, if Hh ligands are present in the tumor microenvironment, then the Hh ligands may activate signaling in the tumor in a reverse paracrine fashion, as exemplified in B-cell lymphoma.[34, 35] Likewise, when activated, these cells can be inhibited by SMO antagonists. Additional evidence of paracrine signaling comes from the high levels of Shh expression in mouse stroma from tumors with high IHH expression, suggesting that the Shh expression may have been induced by Hh signaling in the tumor. Together with IHH in the tumor, Shh in the stroma may participate in the autocrine and paracrine Hh crosstalk between the 2 compartments to promote and amplify the signaling that sustains aberrant Hh activity.

The differential levels of GLI1 and GLI2 expression in osteosarcoma cell lines suggest that GLI1 and GLI2 have nonredundant roles in osteosarcoma. By using SMO agonists, we observed an inverse correlation between endogenous GLI2 expression and Hh pathway induction levels in osteosarcoma cell lines. The finding that Hh pathway induction was solely dependent on GLI2 levels suggests that GLI2 may function as the dominant transcriptional activator and that GLI2 alone may be sufficient to aberrantly activate Hh signaling in a ligand-independent manner. High levels of GLI2 also were reported previously in a small cohort of osteosarcoma samples and cell lines.[31, 32] Noncanonical activation of GLI transcription factors mediated by other pathways (eg, TGFβ, EGF, and MAPK) has been reported.[24, 36] It would be of interest to determine whether these pathways contribute to the high levels of GLI2 observed in the current study. HOS cells that exhibited GLI2 overexpression were sensitive only to GLI inhibition but not to SMO inhibition. Because mutations in the Hh pathway may not be common in osteosarcoma, the presence of GLI2 overexpression may be used to determine which patients have tumors with ligand-independent pathway activation and would not be expected to benefit from SMO antagonist therapies.

IPI-926 specifically blocked signaling in the xenograft models (A and C) that demonstrated ligand-dependent activation in both the tumor and the stroma, thereby disrupting all Hh signaling in the tumor and its microenvironment. Grossly, the antitumor efficacy of IPI-926 was demonstrated in xenograft C but not in xenograft A, although both were responsive to treatment at the molecular level. The number of treatments administered in xenograft A was limited, so it is possible that the treatments may have been terminated before molecular changes could be translated into phenotypic changes. Alternatively, xenograft A may depend on additional pathways for growth, such that inhibition of Hh signaling alone may not be sufficient to abolish tumor growth. Pathway inhibition by IPI-926 appeared to result in a trend toward increased apoptosis in xenograft C and decreased proliferation in xenografts A, B, and C; nevertheless, given the small sample size and the presence of tumor heterogeneity, larger studies will be necessary to confirm these observations.

It is noteworthy that xenografts A and C, which exhibited ligand-dependent activity and sensitivity to IPI-926, were generated from tumors of patients who presented with metastasis at diagnosis. In contrast, xenografts B and D, which exhibited inactive Hh activity and insensitivity to IPI-926, were generated from tumors of patients who presented with localized disease. Because of the limited number of samples in this study, these findings are observational but highlight the need to further investigate the relation of dysregulated Hh signaling to osteosarcoma subtypes and aggressive clinical characteristics.

In this study, we demonstrated that both autocrine and paracrine ligand-dependent activation and ligand-independent activation contribute to aberrant Hh signaling in osteosarcoma and that the sensitivity of osteosarcoma cells to Hh antagonists corresponded with their Hh activity. Because the inhibition of a single pathway is unlikely to have dramatic phenotypic effects in tumors as complex as osteosarcoma, further characterization of dysregulated Hh signaling and its interaction with other signaling pathways in larger sample sizes and appropriate model systems will be necessary to develop multitargeted therapeutic regimens to improve treatment for patients with osteosarcoma.

FUNDING SUPPORT

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

This work was supported by grants from the Canadian Foundation for Innovation and the Ontario Research Fund to J.S.W. and I.L.A. W.W.L. received grants from the Canadian Institutes of Health Research Collaborative Training Program in Molecular Medicine.

CONFLICT OF INTEREST DISCLOSURES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

Veronica Campbell and Karen McGovern are full-time employees of Infinity Pharmaceuticals, Inc., which has a strategic relationship with Purdue Pharmaceutical Products, LP, and Mundipharma International Corporation Ltd.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES