SEARCH

SEARCH BY CITATION

Keywords:

  • RANKL;
  • siRNAs;
  • OSTEOCLASTOGENESIS INHIBITION;
  • BITHERAPY

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

The development of osteosarcoma, the most common malignant primary bone tumor is characterized by a vicious cycle established between tumor proliferation and paratumor osteolysis. This osteolysis is mainly regulated by the receptor activator of nuclear factor κB ligand (RANKL). Preclinical studies have demonstrated that Rankl blockade by soluble receptors is an effective strategy to prevent osteolytic lesions leading to osteosarcoma inhibition. A new therapeutic option could be to directly inhibit Rankl expression by small interfering RNAs (Rkl-siRNAs) and combine these molecules with chemotherapy to counteract the osteosarcoma development more efficiently. An efficient siRNA sequence directed against both mouse and rat mRNAs coding Rankl was first validated in vitro and tested in two models of osteosarcoma: a syngenic osteolytic POS-1 model induced in immunocompetent mice and a xenograft osteocondensant model of rat OSRGA in athymic mice. Intratumor injections of Rankl-directed siRNAs in combination with the cationic liposome RPR209120/DOPE reduced the local and systemic Rankl production and protected bone from paratumor osteolysis. Although Rkl-siRNAs alone had no effect on tumor development in both osteosarcoma models, it significantly blocked tumor progression when combined with ifosfamide compared with chemotherapy alone. Our results indicate that siRNAs could be delivered using cationic liposomes and thereby could inhibit Rankl production in a specific manner in osteosarcoma models. Moreover, the Rankl inhibition mediated by RNA interference strategy improves the therapeutic response of primary osteosarcoma to chemotherapy. © 2011 American Society for Bone and Mineral Research


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

With an estimated incidence of 2 cases per million people per year, osteosarcoma is the most frequent primary bone malignant tumor excluding hematopoietic intraosseous ones1, 2 with a peak incidence at age 18.3 The unifying histologic feature found in all types and subtypes of osteosarcomas is the presence of osteoid tissue produced by the neoplasic cells.4 Since these tumors frequently penetrate and destroy the cortical substance of the bone and extend into the surrounding tissues, an osteolytic activity is often associated with osteosarcoma development. Despite newly devised polychemotherapy courses combined with wide-margin, limb-sparing surgery, osteosarcoma continues to confer a generally poor prognosis in patients who are not responding to chemotherapy or who present lung metastasis at diagnosis (<30% survival rate5).

The biologic problem of tumor development in bone relies on the existence of a vicious cycle between bone resorption and tumor progression.6 Indeed, bone resorption sustains tumor progression by releasing growth factors from the bone matrix. Previous results have shown the benefits of combining chemotherapy with an anti-bone-resorptive agent such as zoledronic acid.7 In a rat osteosarcoma model, bone lesions have been reduced and tumor regrowth has been prevented by the association of zoledronic acid with ifosfamide compared with ifosfamide alone.8 Moreover, lung metastases have been diminished by zoledronic acid alone. These results have provided the rationale for the French clinical trial OS2006 designed to test Zometa (Novartis, Rueil-Malmaison, France) courses combined with conventional chemotherapy and surgery. A favorable risk/benefit ratio is reported for patients treated with bisphosphonates for bone loss associated with cancers. However, the search for specific antiresorptive agents is still necessary because renal dysfunction and osteonecrosis of the jaw can occur with high doses of bisphosphonates9, 10 and because the prolonged administration in pediatric cancers could disturb the growth of long bones.11–13

Receptor activator of nuclear factor κB ligand (RANKL or tumor necrosis factor superfamily member 11, TNFSF11) is the final effector molecule that ultimately stimulates osteoclast differentiation, activation, and survival.14 This osteoclast-promoting cytokine interacts with its receptor RANK (tumor necrosis factor receptor superfamily member 11a, TNFRSF11A) on osteoclast precursors and is negatively regulated by its soluble decoy receptor osteoprotegerin (OPG or tumor necrosis factor receptor superfamily member 11b, TNFRSF11B). For patients with tumor associated with severe osteolysis, the ratio RANKL/OPG is significantly increased both in the tumor environment and in serum,15 suggesting a production of the cytokines by tumor cells. However, it is still a challenge to identify the RANKL-producing cells during both physiologic bone remodeling and bone tumor development.16 Specific neutralization of RANKL has been achieved by either soluble RANK-IgG-Fc,17 recombinant OPG,18 or OPG-like peptidomimetic administrations19 in various preclinical models of osteolytic diseases and in osteosarcoma models.20, 21 A fully human monoclonal antibody directed against RANKL (Denosumab, Amgen, Neuilly sur Seine, France) is currently under investigation in clinical trials to treat patients with bone metastases.22 Thus RANKL is a common therapeutic target for osteolytic diseases associated or not with cancer.23

Another way to specifically target RANKL would be to directly inhibit its production using small interfering RNAs (siRNAs). Since the discovery of RNA interference by Fire in 1998,24 siRNAs have emerged as an undisputed laboratory tool to knock down a gene target, and siRNA-based medicines are now working their way into the clinic (17 clinical trials25 listed at http://clinicaltrials.gov/). Because of the large molecular weight and polyanionic nature, naked siRNAs do not freely cross the cell membrane, and thus delivery systems are required to facilitate their access to their intracellular sites of action.26, 27 In a previous study, we defined an siRNA formulation and an injection protocol efficient to downregulate in vivo the luciferase expression of osteosarcoma cells with no unspecific inhibition of tumor growth.28 Despite the fact that osteosarcoma cells have not yet been identified as a potential source of RANKL, the same protocol was used in this study to investigate the effects of siRNAs directed against mouse or rat Rankl transcripts on osteosarcoma growth and on Rankl production in the tumor environment and the systemic circulation.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

Cell lines and viability analyses

Cell lines

The POS-1 cell line originally isolated from a spontaneous osteosarcoma developed in the mouse C3H/HeN strain was a kind gift from Dr A Kamijo.29 The cells were cultured in RPMI (Roswell Park Memorial Institute, Lonza, Verviers, Belgium) with 10% fetal bovine serum (FBS; Hyclone Perbio, Bezons, France). The OSRGA cell line corresponds to a rat osteosarcoma model originally induced by radiation.30, 31 The cell line designed 293RL was a kind gift from Dr R Josien and derived from the human fetal kidney 293 cell line, which has been modified to express constitutively the mouse Rankl mRNA. OSRGA and 293RL cells were cultured in Dubecco's modified eagle's medium (DMEM, Lonza) supplemented with 10% FBS. The LucF-POS-1 and LucF-OSRGA osteosarcoma cells have been modified to express the firefly luciferase gene, as described previously.28

Mitochondrial activity

Mitochondrial activity was determined by a colorimetric assay using sodium 3'[1-(phenylaminocarbonyl)-3,4-tetrazolium]-bis(4-methoxy-6-nitro-)benzene sulfonic acid hydrate (XTT; Roche Molecular Biochemicals, Mannheim, Germany). Two thousand cells per well were seeded into 96-well plates, and after the culture period, XTT reagent was added to each well and incubated for 5 hours at 37 °C; absorbance then was read at 490 nm using a 96-multiwell microplate reader.

In vitro luciferase activity

Luciferase activity was measured on seeded cells thanks to the Steady-Glo Luciferase Assay System (Promega, Charbonnières, France), which contain lysis buffer and D-luciferin. Then 100 µL of solution was added on cultured cells, and after 5 minutes, the absorbance was read over 10 seconds per well.

Experimental osteosarcoma models

All animals used for in vivo experimentations were housed under pathogen-free conditions at the Experimental Therapy Unit (Faculty of Medicine, Nantes, France) in accordance with the institutional guidelines of the French Ethical Committee (CEEA.PdL.06) and under the supervision of authorized investigators.

Tumor induction

The mice were anesthetized by inhalation of an isoflurane-air mixture (2%, 0.2 L/min) prior to any surgical procedures. Four-week-old male C3H/HeN mice or 4-week-old male nude NMRI mice (Elevages Janvier, Le Genest Saint Isle, France) were used for POS-1 or OSRGA tumor induction, respectively. A 5-mm section was performed in naive mice to open the muscle along the tibia. Tumor fragments (2 × 2 × 2 mm3) were transplanted into naive mice in close contact with the tibial diaphysis, and then muscular and subcutaneous wounds were sutured. The tumor volume was quantified by measuring two perpendicular diameters with a caliper and calculated with the formula: (l2 × L)/2, where l is the smallest and L the largest diameter.

Samples

At euthanasia, blood samples were drawn by intracardiac puncture under deep anesthesia just before euthanasia. Lungs were systematically observed for macroscopic tumors at necropsy. Tumor and bone samples were collected for histologic analyses.

siRNAs

The customer-specified oligonucleotide duplexes (GAU CUC UAA CAU GAC GUU Att annealed with UAA CGU CAU GUU AGA GAU Ctt and GAU GGC UUC UAU UAC CUG Utt annealed with ACA GGU AAU AGA AGC CAU Ctt) target Mus musculus Rankl mRNA (NM_011613) at start positions 749 and 795, respectively. A control siRNA duplex (UUC UCC GAA CGU GUC ACG Utt annealed with ACG UGA CAC GUU CGG AGA Att), which did not show significant homology with any mouse or rat mRNA sequence according to BLAST database searches, was used as a negative control and designed CT-siRNA. A positive control siRNA duplex targeting the luciferase expression was designed: LucF-siRNA (sequences in Rousseau and colleagues, 2010). All siRNAs were purchased from Eurogentec (Angers, France).

Transfection with INTERFERin

Rkl-749 and Rkl-795 siRNAs were transfected at 10 nM with 3 µL of INTERFERin (Ozyme, Saint Quentin Yvelines, France) in tripliclate.

Transfection with cationic liposome

0.3 µg of siRNAs were premixed with DNA (a plasmid containing no eucaryotic expression cassette) 1:1 and formulated into lipoplexes with cationic liposome RPR209120/DOPE at a ratio of 6 nmol of cationic lipid per microgram of nucleic acid. Fifty percent confluent cells were transfected in triplicate.

In vivo injections

Groups of 8 mice transplanted with POS-1 or OSRGA osteosarcoma fragments were assigned as vehicle (NaCl injection), CT-siRNA, or Rkl-siRNA. Based on preliminary results, 2.5 µg of siRNA combined with cationic liposomes at a ratio of 6 nmol of cationic lipid per microgram of nucleic acid was injected into the tumor mass three times a week.28 The injection started at the diaphysis bone contact and moved through the tumor mass by a continuous and slow pressure. The combined chemotherapeutic cure consisted in three consecutive intraperitoneal injections of ifosfamide (ASTA Medica Laboratories, Mérignac, France) at 24-hour intervals per week: three injections of 60 mg/kg over 3 weeks and 6 mg/kg over 2 weeks for the POS-1 and OSRGA models, respectively.

Rankl detection

Western blotting

Transfected cells were lysed with a lysis buffer (150 mM NaCl, 5% Tris [pH 7.4], 1% Nonidet P-40, 0.25% Na deoxycholate, 1 mM Na3VO4, 0.5 mM phenylmethylsulfonyl fluoride, 10 µg/mL of leupeptin, and 10 µg/mL of aprotinin). Total protein concentrations were determined using the bicinchoninic acid based method (BCA, Sigma, St Quentin-Fallavier, France). Twenty µg of total cell lysate proteins was run on an SDS–polyacrylamide gel electrophoresis column and then transferred electrophoretically to an Immobilon-P membrane (Millipore Corporation, Billerica, MA, USA). The membrane was blotted with mouse antibodies to Rankl (A5060; Sigma-Aldrich, St Louis, MO, USA), and rabbit anti-mouse actin was used as a loading control (40940; Active Motif, Rixensart, Belgium) in PBS, 0.05% Tween, and 3% bovine serum albumin (Sigma). The membrane was washed and probed with a secondary antibody coupled with horseradish peroxidase. Antibody binding was visualized with the enhanced chemoluminescence system (SuperSignal West Dura Kit, ThermoScientific, Rockford, IL, USA).

Enzyme-linked immunosorbent assay

Quantifications of Rankl in serum isolated from blood samples were performed with the DuoSet Mouse TRANCE/TNFS11 Kit (R&D Systems, Lille, France).

Immunohistochemistry

Tumor tissues were fixed in 4% buffered formaldehyde over 1 week before embedding in paraffin. Then 3-µm-thick sections were deparaffinized in OTTIX PLUS (xylene and alcohol substitute; DiaPath, Martinengo, Italy) and hydrated through graded alcohols. All the following steps were performed using the Autostainer 360-2D (Microm Microtech, Francheville, France). Endogenous peroxidase was removed using 3% H2O2 in water for 10 minutes. Tissues were blocked with 4% bovine serum albumin for 15 minutes. Sections were incubated with diluted goat anti-RANKL(N19) primary antibody (1:30; Santa Cruz Biotechnology, Heidelberg, Germany) for 1 hour. Sections were incubated again at room temperature with biotin-conjugated anti-goat secondary antibody (Vector Laboratories, Burlingame, CA, USA) for 30 minutes. After addition of streptavidin-conjugated horseradish peroxidase for 30 minutes, immunodetection was analyzed using 3-amino-9-ethylcarbazole (AEC; Sigma-Aldrich, St Louis, MO, USA) and counterstained with Mayer hematoxylin. All pictures were assessed by light microscopy using a DMRXA microscope (Leica, Nussloch, Germany).

Flow cytometry

Enhanced green fluorescent protein (EGFP)-expressing POS-1 tumors were first lacerated in PBS and then treated with 0.1% collagenase A and 0.3% elastase (both from Roche Applied Science, Meylan, France). After filtration, cells were washed with PBS 0.1% BSA (Sigma) and FcR blocking reagent (BD Biosciences, Le Pont de Claix, France). Two-hundred thousand cells were finally incubated with anti-Rankl antibody (clone IK22-5; BD Biosciences) and washed before analysis with the FC500 cytometer (Beckman Coulter, Villepinte, France).

Immune response

To assess the possible immune system activation through lipoplex (siRNA plus cargo plus cationic liposome) injections in mice, the interferon-γ concentration was measured on 50 µL of blood serum with the Mouse IFN-γ ELISA Kit II (BD Biosciences). Sera from mice that have received modified vaccinia virus Ankara (MVA) empty vectors known to induce an interferon-γ response32 were used as positive controls. Sera were collected 72 hours after the last siRNA or MVA vector injection.

Bone parameter analyses

Microscanner analyses

Bone architecture was analyzed for tumor-bearing and naive tibias using the high-resolution SkyScan-1072 X-ray micro–computed tomographic (µCT) system (SkyScan, Kartuizersweg, Belgium). All tibias were scanned at necropsy using the same parameters: 18 µm of pixel size, 50 kV, 0.5-mm Al filter, and 0.8 degree of rotation step. 3D reconstructions and analysis of bone parameters were performed using ANT and CTan software (Skyscan), respectively. Calculation of tibia relative bone volume (bone volume/total volume [BV/TV]) following 3D morphometric parameters (bone ASBMR nomenclature) was performed on 8 mm of tibia length from the perone insertion. This area corresponds to bone in close contact with osteosarcoma tumors and excludes trabecular bone.

TRAP staining

Tumor-bearing tibias were fixed in 10% buffered formaldehyde and decalcified by continue electrolysis over 72 hours before embedding in paraffin. Transversal 3-µm-thick sections obtained from the tibia area were stained for tartrate-resistant acid phosphatase (TRAP) to identify osteoclasts by 1-hour incubation in a 1 mg/mL naphthol AS-TR phosphate, 60 mmol/L N,N-dimethylformamide, 100 mmol/L sodium tartrate, and 1 mg/mL Fast red TR salt solution (all from Sigma Chemical Co.) and counterstained with hematoxylin.

Statistical analysis

All analyses were performed using GraphPad InStat Version 3.02 software (GraphPad Software, La Jolla, CA, USA). In vitro experimentation results were analyzed with the unpaired nonparametric Mann-Whitney U test using two-tailed p values. Results are given as mean ± SD, and results with p < 0.05 were considered significant. Concerning in vivo experiments, multivariate approaches based on nonparametric t test one-tailed p values were used. Since no significant differences were observed between the CT-siRNA and untreated tumors, results are shown only between Rkl-siRNA and CT-siRNA groups as mean ± SEM, with p < 0.05 considered significant.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

Identification of a highly efficient siRNA sequence to reduce Rankl production in vitro

Following the guideline of Reynolds and colleagues,33 two customer-specified siRNAs were designed to inhibit both mouse and rat Rankl gene expression by targeting two 19-nucleotide mRNA sequences starting at positions 749 and 795 in Mus musculus gene (NM_011613). The downregulation of Rankl by these siRNAs was not tested in vitro on the mouse and rat osteosarcoma cells because these cells do not express Rankl under basic culture conditions.20 They were transiently transfected with INTERFERin into 293RL cells that constitutively express the mouse full-length Rankl mRNA. Rkl-749 siRNA-transfected 293RL cells showed a slight reduction in Rankl detection by Western blotting compared with untransfected or CT-siRNA-transfected cells that have no target in mouse or human cells. In contrast, Rankl was almost undetectable when cells were transfected with the Rkl-795 siRNAs (Fig. 1A, left panel). Also, this more powerfull Rkl-795 siRNA (next named Rkl-siRNA) was efficient when formulated in lipoplexes with the cationic liposome RPR209120/DOPE (Fig. 1A, right panel).

thumbnail image

Figure 1. In vitro validation of Rankl-directed siRNAs formulated in lipoplexes. (A) HEK 293 cells modified to express mouse Rankl were transfected with CT-, Rkl-749, or -795 siRNAs combined with INTERFERin or RPR209120/DOPE. Rankl was detected by Western blotting on cell lysates performed 48 hours after transfection and compared with untransfected cells (CT). (B) Osteosarcoma cell lines modified to express luciferase (LucF-OSRGA and LucF-POS-1) were transfected with LucF- or Rkl-795 siRNAs combined with RPR209120/DOPE. The cell metabolism was analysed 48 hours after transfection by measuring the luciferase activity and the mitochondrial activity in a XTT assay (absorbance at 490 nm). These experiments were performed three times.

Download figure to PowerPoint

Despite the fact that tumor cells do not express Rankl under basic culture condition, we checked whether the Rkl-siRNA lipoplexes have a direct effect on the viability of the mouse LucF-POS-1 and rat LucF-OSRGA osteosarcoma cells by measuring luciferase and mitochondrial activities. LucF-OSRGA and LucF-POS-1 cells were efficiently transfected because the luciferase activity was reduced by 60% and 35%, respectively, with siRNAs directed against luciferase mRNA (Fig. 1B). Using the same transfection conditions, the Rkl-siRNA lipoplexes did not change the luciferase activity (Fig. 1B, left panel) nor the mitochondrial activity (Fig. 1B, right panel). Moreover, cell counting after trypan blue staining (data not shown) confirmed that in vitro proliferation of mouse and rat osteosarcoma cells was not affected by Rkl-siRNAs lipoplex transfections.

Rankl-directed siRNAs injected alone have no effect on tumor growth but reduce osteosarcoma-induced osteolysis

When rat OSRGA tumor is induced in a nude mouse, Rankl is detected in blood serum. In contrast, the Rankl level remains low in serum of C3H/HeN mouse bearing POS-1 tumor. In a first attempt, we took advantage of this OSRGA model to test the silencing efficiency of Rkl-siRNAs by following Rankl levels in blood serum. With siRNAs formulated in lipoplexes injections in OSRGA tumors at the time of their detection, the Rkl-siRNA-injected mice showed a significant decrease in systemic Rankl concentration of 48.5% compared with the CT-siRNA group after 3 weeks of treatment (Fig. 2A). However, the Rankl downregulation was not associated with OSRGA tumor growth modulation. Indeed, the progression of tumor volume was similar between the Rkl-siRNA and the CT-siRNA groups (Fig. 2B).

thumbnail image

Figure 2. In vivo validation of Rkl-siRNAs formulated in lipoplexes. (A) Rankl concentration was measured by ELISA assay on blood samples from OSRGA tumor-bearing mice after 3 weeks of treatment: individual measures are presented as dots, and mean concentrations (line) is shown for the groups treated with vehicle, CT-siRNA, or Rkl-siRNA lipoplexes. a and NS indicate a significant difference (p < 0.05) or a nonsignificant difference, respectively. (B) Mean tumor volumes (line) and individual measures (dots) are shown for OSRGA tumor-bearing mice treated with CT-siRNA or Rkl-siRNA lipoplexes. (C) One representative tibia is shown for a mouse bearing no tumor (naive) or a POS-1 tumor-bearing mouse of each treated group (vehicle, CT-siRNA, and Rkl-siRNA). Microscanner analyses were performed to evaluate the specific bone volume (BV/TV). (D) Interferon-γ concentrations were measured by ELISA assay on blood samples from POS-1 tumor-bearing mice treated with vehicle, CT-siRNA, or Rkl-siRNA lipoplexes (n = 8) and compared with mice (n = 3) injected with modified vaccinia virus Ankara empty plasmid (MVA). Individual measures are presented as dots and mean concentrations as lines. This study was repeated another time.

Download figure to PowerPoint

Anti-bone-resorptive effects are difficult to quantify in the OSRGA model because osteolysis is masked by strong osteogenic lesions on tumor-bearing tibias. To test whether Rankl downregulation by Rkl-siRNAs could have a specific inhibitory effect on bone resorption, we next injected siRNAs with the same protocol in the mouse osteolytic POS-1 model. Similar to previous observations, the Rkl-siRNA injections alone did not modify tumor growth (data not shown). However, the microarchitecture analysis of a representative tibia from each group showed a protective effect of Rkl-siRNAs against the tumor-induced bone lesions compared with CT-siRNA or vehicle-treated groups (BV/TV of 54.3% versus 48.4% and 47.9%, respectively; Fig. 2C).

Additionally, the production of interferon-γ, which could be due to an immune innate and/or adaptive response and corresponds to an unspecific off-target effect of siRNAs, was quantified in the blood sera of mice. As compared with the vehicle group, we observed a 10-fold increase of interferon response associated to vectorized siRNA injections in C3H/HeN mice bearing POS-1 tumor, whereas a 150-fold increase was observed in same strain mice of injected with MVA empty vectors (Fig. 2D). The same results were noticed for the OSRGA model induced in nude mice with a 3-fold increase of IFN concentration after vectorized siRNA injections. This increase was 35-fold reduced compared with the one observed in animals that received MVA vectors (data not shown).

In an osteolytic osteosarcoma model, Rankl-directed siRNAs enhanced tumor response to chemotherapy and slowed tumor regrowth after chemotherapy arrest

Because the Rkl-siRNAs enabled us to protect bone from osteolysis induced by osteosarcoma, we next asked whether the Rkl-siRNAs could enhance the effects of chemotherapy. Therefore, new experiments using the POS-1 osteolytic osteosarcoma model were conducted combining the Rkl-siRNAs with the chemotherapeutic agent ifosfamide, which was already shown to slow tumor progression in the same osteosarcoma model for approximately 50% of patients.28

The results presented in Fig. 3A clearly demonstrate a significant decrease in the mean tumor volume in the three groups of mice treated with ifosfamide (ifosfamide associated with either vehicle, CT-siRNA, or Rkl-siRNA) as compared with the control group (vehicle) as early as day 9 after clinical detection of tumor (p < 0.05), when two courses of ifosfamide have been administered on days 1 and 8. On day 14, the mean tumor volume of mice treated with ifosfamide plus Rkl-siRNAs was significantly lower than that of mice treated with either ifosfamide alone or ifosfamide plus CT-siRNAs (p < 0.05; Fig. 3A). This result indicated an additive effect of Rkl-siRNA injections to chemotherapy response compared with CT-siRNA injections. After the last ifosfamide course, the tumors of mice treated with vehicle or CT-siRNAs grew faster than those of mice treated with Rkl-siRNAs until the end of the experiment (p < 0.05; Fig. 3A). This slower tumor progression induced by the Rkl-siRNA injections is evidenced in Fig. 3B, which presents the averages of relative tumor progression in each group for each animal between days 13 and 25 corresponding to tumor regrowth.

thumbnail image

Figure 3. Rkl-siRNA injections combined with ifosfamide in the osteolytic POS-1 osteosarcoma model. C3H/HeN mice were transplantated with POS-1 tumor fragment and devided in four groups; vehicle (NaCl injections) associated or not with ifosfamide treatment, CT-siRNA, and Rkl-siRNA groups were both combined with the chemotherapeutic agent. Injections of vehicle or siRNAs were started at tumor clinical detection that corresponds to day 0 for each mouse. (A) The mean tumor volume evolution of each group after three treatments with ifosfamide (arrows) is shown. (B) Percentage of tumor regrowth was calculated with the following formula: (Tumor volume at day 25 – tumor volume at day 13)/tumor volume at day 13. a indicates significant differences between CT- and Rkl-siRNA groups (p < 0.05) for A and B. (C) Microscanner analyses and BV/TV percentage calculations were performed. Mean BV/TV%, SD, and animal number (n) are indicated for each treated group divided into three groups of tumor volume values. This experiment was performed two times.

Download figure to PowerPoint

The effects of Rkl-siRNAs were analyzed at the bone level by microscanner analyses (Fig. 3C). Bones in contact with a tumor with volume greater than 1500 mm3 were associated with a low specific bone volume compared with tibias of naive mice: mean BV/TV of 44%, 41.5%, or 51.4%, respectively, for the groups treated with vehicle or ifosfamide or ifosfamide plus CT-siRNAs versus 66.1% for tibias without tumor. In contrast, small tumors (volume < 200 mm3) were associated with weak bone lesions with specific bone volumes identical to or slightly increased compared with tibias of naive mice (mean BV/TV of 67.6%, 69.1%, or 69.5%, respectively, for the groups treated with vehicle, ifosfamide, or ifosfamide plus CT-siRNAs). Intermediate specific bone volumes (57.9% to 66.8%) were calculated for tumors between 200 and 1500 mm3. The group treated with ifosfamide plus Rkl-siRNAs did not present tumors greater than 1500 mm3 and showed the highest specific bone volumes: 72.7% and 74.3% for tumor volumes of 200 to 1500 mm3 or less than 200 mm3, respectively. Overall, these values highlighted a marked osteolysis for mice bearing big tumors and demonstrated the efficiency of Rkl-siRNAs toward bone degradation, as evidenced by the more elevated BV/TV ratio in this group compared with all other groups, even with the naive mice.

This prevention of paratumor osteolytic bone lesions was associated with a decrease in osteoclast-like cells detected by TRAP staining for Rkl-siRNA-treated tumors. This decrease of osteoclast number was observed only in close contact with the tumor at the tibial diaphysis that corresponds to the site of lipoplex injection, whereas numerous osteoclasts were still detected by TRAP staining wihin the epiphyses near growth plates (Fig. 4A).

thumbnail image

Figure 4. TRAP and Rankl detection after Rkl-siRNA injections in POS-1 osteosarcomas. One representative photomicrograph of histology analyses is shown for the group of POS-1 tumor-bearing mice treated with ifosfamide plus CT-siRNAs or the one treated with ifosfamide plus Rkl-siRNAs. (A) TRAP staining was performed to identify osteoclasts at the interface between tumor and cortical bone (B). (B) Immunodetection of Rankl was performed in tumor tissues. These parameters were analyzed for two times on two different studies. (C) Detection of EGFP- and Rankl-expressing POS-1 cells, which were dissociated from tumors and analyzed by flow cytometry following incubation with anti-Rankl antibody combined with PE (right panel). Left panel shows FL1 and FL2 negative controls using POS-1 tumor cells that do not express EGFP and that were incubated with the isotype control antibody combined with PE. The middle panel shows EGFP-POS-1 cells incubated with the isotype control antibody. Numbers indicate the percentage of cells in each quadrant.

Download figure to PowerPoint

The immunohistochemistry detection of Rankl revealed positive cells within the tumor samples of mice treated with ifosfamide plus CT-siRNAs, especially at the interface of necrotic areas and proliferative areas of tumors (Fig. 4B). In contrast, for mice treated with ifosfamide plus Rkl-siRNAs, the Rankl+ cells were not detected within small tumors (volume < 200 mm3) and were less abundant and more dispersed within larger tumors than their counterparts in mice treated with ifosfamide plus CT-siRNAs. Rankl+ cells detected by immunohistochemistry are likely corresponding to POS-1 cells. Indeed, after cell dissociation, EGFP-expressing POS-1 tumor, and flow cytometric analyses, up to 24% of POS-1 cells were stained with Rankl-directed antibody (Fig. 4C). This result demonstrated that despite osteosarcoma, cells do not produce in vitro Rankl; in a tumor context, they are able to do it. Moreover, the intratumor injection of rhodamine-labeled lipoplexes demonstrated that 27% of cells in the tumor mass were efficiently transfected 24 hours after injection (data not shown). Taken together these data suggested that Rkl-siRNA injections likely have downregulated Rankl production within osteosarcoma cells and have inhibited the osteoclast differentiation at the bone-tumor interface.

In an osteogenic osteosarcoma model, Rankl-directed siRNAs induced tumor regrowth delay after cessation of chemotherapy

We next investigated the effect of Rkl-siRNAs combined with ifosfamide in the rat OSRGA model of osteosarcoma, which induces osteolytic lesions together with predominant osteogenic ones, as observed in patients. As shown in the Fig. 5A, the courses of ifosfamide were highly efficient to stop the tumor development in the three treated groups of mice (ifosfamide associated with either vehicle, CT-siRNAs, or Rkl-siRNAs) compared with the untreated group (vehicle) as early as day 9 after clinical detection of tumor. Thirteen days after the last ifosfamide injection, the animals with prolonged injections of Rkl-siRNAs showed smaller tumor volumes than those with prolonged injections of vehicle or CT siRNAs (p < 0.05). Moreover, the Rkl-siRNA-treated tumors grew significantly slower from day 42 until day 54 compared with the two control groups treated previously with ifosfamide (Fig. 5A).

thumbnail image

Figure 5. Rkl-siRNA injections combined with ifosfamide in the osteocondensent OSRGA osteosarcoma model. Nude mice were transplantated with an OSRGA tumor fragment and devided in four groups: vehicle associated or not with ifosfamide, CT-siRNA, and Rkl-siRNA groups plus ifosfamide. Injections of vehicle or siRNAs were started at tumor clinical detection, which corresponds to day 0 for each mouse. (A) Mean tumor volume evolution of each group after two courses of ifosfamide (arrows) is shown. (B) Systemic Rankl concentration was evaluated on blood sera by ELISA assay. a shows statistically differences between CT- and Rkl-siRNA groups (p < 0.05) for panels A and B. (C) Microscanner analyses and BV/TV percentage calculations were performed. Mean BV/TV%, SD, and animal number (n) are presented for each treated group divided into two groups of tumor volume values and for naive bone. This experiment was performed two times.

Download figure to PowerPoint

Moreover, at the end of the experiment, all mice (n = 8) treated with ifosfamide plus Rkl-siRNAs showed lower Rankl concentrations (<6 pg/mL) than mice treated with ifosfamide plus vehicle or plus CT-siRNAs (Fig. 5B). These observations could result both from the downregulation of Rankl expression by RNA interference and from the moderate development of tumors in the group of mice treated with Rkl-siRNAs.

Despite visible bone remodeling induced by the tumor development, the BV/TV calculations showed no difference for untreated tumors (50.6% and 51.5% for tumor greater than or less than 900 mm3, respectively) compared with naive bone (53%; Fig. 5C). These results highlight the difficulty in discriminating the cortical bone from the ectopic bone formation by OSRGA tumor itself. Ifosfamide treatments may have induced a slight gain in bone (55.4% and 57.3% for ifosfamide alone 54% and 53.5% for ifosfamide plus CT-siRNAs injections). The tibias of animals treated with ifosfamide plus Rkl-siRNAs seemed to exhibit a greater BV/TV of 63.3% for tumors greater than 900 mm3 and 59.8% for small ones.

Osteoclast-like cells were detected by TRAP staining for tibias of mice treated with ifosfamide plus Rkl-siRNAs or CT-siRNAs. Osteoclast number and size were clearly reduced at the site of Rkl-siRNA injection (diaphysis) as well as at the growth plates (data not shown). Therefore, by inhibiting osteoclast differentiation, Rkl-siRNA formulated in lipoplexes enabled to contain osteosarcoma progression even when the bone formation is predominant compared with the osteolysis.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

The pivotal role played by the cytokine RANKL in the regulation of bone resorption makes it an attractive therapeutic target in bone diseases with osteolytic lesions, including those from tumor origin.34–37 In bone tumors, blocking RANKL activity leads to indirect inhibition of tumor cell growth by limiting the release of growth factors stored in the bone matrix.6 This strategy had been extended to osteosarcomas, which are not only associated with osteolytic lesions but also are first defined as a malignant bone tumor with osteogenic properties.38–40 Indeed, our group had demonstrated at the preclinical level the benefit effect of RANKL decoy receptors not only in osteolytic but also in osteogenic osteosarcomas.20, 21 In this study, we confirmed the efficiency of targeting RANKL factor in the tumor microenvironment as a therapeutic strategy for both types of osteosarcomas.

The inhibition of Rankl by a specifically designed siRNA sequence formulated in lipoplexes was associated with a decrease in Rankl function, as revealed by the decrease in osteolytic lesions and osteoclast number. POS-1 tumors developing in mice produce Rankl locally, as demonstrated by Rankl immunohistochemical and flow cytometric analyses but not at levels high enough to be detectable in mouse serum. Thus no effect of Rkl-siRNAs can be detected on serum Rankl for POS-1-bearing mice. In contrast, OSRGA tumors-induced a serum Rankl increase that was reduced by Rkl-siRNA injections compared with control siRNAs. For both models, Rkl-siRNAs reduced osteoclast differentiation at the bone-tumor interface, which was the injection site (diaphysis), and also at the growth plates (epiphyses) for the OSRGA model only.

The identification of efficient siRNAs targeting proresorptive Rankl in bone tumor models supports the idea that siRNAs are promising therapeutic molecules presenting some advantages of chemical molecules (fast synthesis) and others of proteins or antibodies (strong specificity). Their pharmaceutical development could be faster than the strategies that led to the development of other anti-bone-resorptive drugs such as Zometa and Denosumab.25, 41

Because spontaneous lung metastases were not detected in all untreated and treated mice, this study cannot address the potential therapeutic effect of Rkl-siRNAs on tumor dissemination. Endo-Munoz and colleagues have reported that the loss of osteoclasts induced by zoledronic acid was correlated with an increase in osteosarcoma lung metastases in two orthotopic and xenogenic models.42 In contrast, we didn't observe an activated metastasic process after osteoclast decrease in the tumor microenvironment by Rkl-siRNA treatments. This apparent discrepancy can result from the difference between the models used in both studies. It will be of high interest to compare the effects of zoledronic acid and Rkl-siRNA on osteosarcoma metastases in an appropriate model with lung dissemination.

Rkl-siRNAs alone had no effect on osteosarcoma proliferation despite the fact that they were efficient to decrease Rankl production and to prevent paratumor osteolysis. In contrast, the osteolysis inhibition mediated by OPG gene transfer has been effective in preventing the development of POS-1 tumors and in inhibiting OSRGA tumor growth even in the absence of chemotherapy.20 Likely, the injection of Rkl-siRNAs in higher quantities could provide antitumor effects, as observed in two independent experiments where a decrease in osteosarcoma growth was measured using combined Rkl-siRNAs alone compared with untreated tumors (data not shown). However, this antitumor effect was due in part to unspecific off-target effects of siRNAs, as indicated by a slight (but not significant) decrease in osteosarcoma growth that was observed in the CT-siRNA-treated group compared with the untreated group. These unspecific off-target effects of siRNAs may result from activation of interferon responses either through nucleic acid recognition by toll-like receptor (TLR) in immune cells or through protein kinase R and RNase L activation in transfected cells, including tumor cells.43–45 Furthermore, the association of siRNA with cationic liposome, which enables cell transfection, also could be associated with toxicity, especially by modifying the pathway of siRNA uptake and its subsequent binding to endosomal TLR.46, 47 Additionnally, a high dosage of exogeneous synthetic siRNAs may interfere with the endogeneous pathway of RNAi mediated by micro-RNAs.

In this study, we measured interferon-γ at the end of the experiment and did not observe a high response, contrary to what Ma and colleagues observed on injection of siRNA lipoplexes.46 This discrepancy could be explained by the use of a different cationic lipid. In our case, it would be of interest to measure interferon-α/β, interleukin 6, or tumor necrosis factor α immediately following lipoplex injection as siRNA, and lipoplexes have been shown to be potent activators of the mammalian innate immune system.47

Anti-bone-resorption molecules such as biphosphonates and RANKL inhibitors are actively developed as an effective adjuvant therapy to prevent and reduce skeletal-related events (SREs).22 In this context, osteosarcoma patients can receive conventional chemotherapy associated with intravenous biphosphonate injections such as zoledronic acid in the current phase 3 French clinical trial (NCT00470223) or pamidronate in the complete phase 2 American trial48 (NCT00072306, http://clinicaltrials.gov). This association of pamidronate and chemotherapy showed an efficient decrease in paratumor osteolysis for treated patients.49 However, previous clinical studies on bone metastases from breast or prostate cancer reported SRE recurrence despite continuous injection of zoledronic acid,50 which is considered to be the most efficient biphosphonate.51 Moreover, its high administration is associated with a nephrotoxicity that increases with extended treatments.52, 53 RANKL inhibitors prevented SREs similarly to zoledronic acid, but in patients who had poor response to biphosphonates, the Denosumab was more efficient than zoledronic acid injections.50, 54, 55

Moreover, the development of such combined therapies between RANKL inhibitors and cytotoxic agents also could reduce the dose of chemotherapy that is associated with undesired side effects (eg, tissue toxicity, resistance, and relapse). Indeed, in preclinical studies, OPG administration in a model of prostate cancer bone metastases improved tumor response to docetaxel, known to induce a cell cycle arrest and apoptosis of tumor cells.56 In a human bone metastasis model, OPG also enhanced the therapeutic effect of an antibody targeting the epidermal growth factor receptor implicated in cell growth and death.57 In our study, the association between Rkl-siRNA injections and ifosfamide courses showed a synergistic effect on blocking tumor progression and delaying tumor relapse. In this context, the association of an anti-bone-resorption siRNA and an antitumor agent offers new and interesting opportunities to develop bitherapy protocols. Few preclinical studies performed on osteosarcomas have validated RNA interference molecules that target genes implicated in tumor metabolism such as the glucose transporter protein 158 in cell cycle with inhibition of stathmin production59 or in apoptosis by targeting Bcl-2.60 Additionally, three siRNAs targeting three different transcripts were formulated with the cationic liposome RPR209120/DOPE and enabled significant improvement in the clinical pattern of murin arthritis compared with each vectorized siRNA alone.27 From these data, we now can consider combining antitumoral siRNAs with antiresorptive ones for multitargeting in a context of tumor development at bone sites. This will offer new opportunities for patients who do not respond to conventional chemotherapeutic protocols.

Disclosures

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

All the authors state that they have no conflicts of interest.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

We thank Dr Georges Vassaux and Lucile Tran from the U948 unit for providing serum of vaccine mice with MVA vectors. Moreover, this study was supported by the Région des Pays de la Loire (JG/ND/RECH N 660, fellowship for JR) and the Agence Nationale de la Recherche 2007 “Pathophysiology of Human Diseases” Project N R07196NS.

Author's roles: Study design: VT, FR, and DH. Study conduct: JR and FL. Technical assistance: SB, RB, JC, and JA. Experts in liposomes and revising manuscript: VE and DS.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References