Background
Osteosarcoma is the most common primary malignant bone tumor with a peak incidence in childhood and adolescence frequently occurring at sites of rapid bone growth. The long-term survival of patients with osteosarcoma has improved from 10 to 20 % to nearly 80 % within the last 25 years, due to the use of neoadjuvant chemotherapy [
1]. However, this plateau has not changed for more than 15 years [
2]. For patients with metastatic disease, the outcomes are distinctly worse, with less than 30 % survival at 5 years [
3]. Furthermore some patients do not respond to chemotherapy and others exhibit features of multidrug resistance (MDR), probably due to overexpression of P-glycoprotein (ABCB1) [
4]. Consequently, there is still an urgent demand for new and more effective therapeutic strategies.
The erbB family receptor tyrosine kinases play an important role in the control of cell cycle, proliferation, and migration of normal and cancer cells. Expression of erbB molecules, especially the epidermal growth factor receptor (EGFR, erbB1), has been described in osteosarcomas [
5]. This observation opens the possibility that EGFR might be a feasible therapeutic target in osteosarcoma although activating EGFR mutations are widely missing in this aggressive primary bone tumor [
5,
6]. EGFR-targeted therapies (monoclonal antibodies or small molecule TK inhibitors) have been already clinically approved for several tumors including colorectal, non-small cell lung, head and neck, as well as pancreatic cancer [
7‐
9]. Gefitinib is a selective EGFR tyrosine kinase inhibitor currently utilized for the treatment of patients with non-small cell lung cancer harbouring activating EGFR mutations [
10‐
12]. In the present study, we investigated the impact of EGFR inhibition on osteosarcoma cell behaviour and its interaction with chemotherapy response.
Methods
Reagents
Gefitinib (ZD1839, Iressa®, 4-quinazolinamine, N-(3-chloro-4-flurophenyl)-7-methoxy-6-(3-(4-morpholinyl) propoxy), a selective EGFR inhibitor, was kindly provided by AstraZeneca, and dissolved in sterile dimethyl sulphoxide (DMSO). Solutions were freshly prepared before use. As standard chemotherapeutic agents in osteosarcoma, doxorubicin, methotrexate (MTX) and cisplatin were used (Sigma, Vienna, Austria). Stock solutions were prepared for gefitinib at 10 mM in DMSO, doxorubicin at 3.5 mM in saline, cisplatin at 4 mM in dimethylformamide and MTX in a minimal amount of 1 M NaOH followed by saline to 200 mM.
Cell cultures
MG-63, HOS, Saos-2 and U-2 OS osteosarcoma cell lines were obtained from the American Type Culture Collection (Manassas, VA). Primary osteosarcoma cell lines OS-10, OS-9, SARG and IOR-MOS were generously supplied by Dr. K. Scotlandi from the Instituti Ortopedici Rizzoli (Bologna, Italy) [
13,
14]. The HL-NG cell line was established from a fibroblastic osteosarcoma surgery specimen at the Institute of Cancer Research, Vienna. Calu-3, an EGFR-overexpressing non-small cell lung cancer cell line used as positive control, was obtained from ATCC. Histological subtype of the original tumors and culture media are indicated in Additional file
1: Table S1. Cells were cultured in growth media supplemented with 10 % fetal calf serum (FCS) at 37 °C in a 5 % CO
2 incubator. The cell lines were authenticated in all cases by array comparative genomic hybridization (Agilent, 44 k human whole genome DNA arrays) as published [
15] and/or short tandem repeat (STR) fingerprinting before the start of this study.
Cell growth and viability assays
Cells were plated (2x10
4 cells/mL) in 100 μL per well in 96-well plates and allowed to attach for 24 h. Drugs were added in another 100 μL growth medium and cells exposed for 72 h. The proportion of viable cells was determined by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium assay (MTT) following the manufacturer’s recommendations (EZ4U, Biomedica, Vienna, Austria). Cytotoxicity was expressed as IC
50 values calculated from full dose–response curves. The interaction between the activities of combined drugs is expressed by the combination index (CI) as published by Heffeter et al. [
16] using CalcuSyn software (Biosoft, Ferguson, MO). CI < 0.9, CI = 0.9–1.2 or CI >1.2 represent synergism, additive effects and antagonism, respectively.
Apoptosis and cell cycle analyses
Induction of cell death was followed by staining of living osteosarcoma cell cultures under different treatment and serum conditions with Hoechst 33258 and propidium iodide (PI) as published [
17]. Hoechst dye enters living cells and allows detection of chromatin condensation due to either mitosis or apoptosis. In contrast, PI only stains dead cells due to necrosis or apoptosis. Assays were performed in 24-well plates in duplicate and 4 optical fields per well were evaluated after 24 h by microscopic counting of apoptotic cells. Cell cycle distribution was analysed after 24 h exposure to gefitinib by PI staining of ethanol-fixed cells followed by FACS analysis as published previously [
17,
18].
Clonogenic assay
103 cells per well were seeded into six-well plates. Following 24 h recovery, cells were treated with MTX, doxorubicin, cisplatin and gefitinib. At day 7 of exposure, cells were washed twice with PBS, fixed with methanol at −20 °C and stained with crystal violet. The number of colonies containing at least 100 single cells was determined by counting microscopically using a Leica DMIL (Leica, Solms, Germany).
Protein isolation and Western blotting
Total protein and membrane protein-enriched fractions were extracted and processed for Western blotting as described [
19,
20] using the following primary antibodies: pEGFR(Tyr1068), EGFR, phospho-S6 ribosomal protein (Ser240/244), S6, phospho-p44/42 MAP kinase ERK (Thr202/Tyr204), p44/42 MAP kinase ERK, pAKT (Ser473), AKT, pGSK3β (Ser9), GSK3β, (all polyclonal rabbit contained in the respective sampler kits from Cell Signaling Technology, Beverly, MA),
ß-actin monoclonal mouse AC-15 (Sigma).
RNA isolation
Total RNA was isolated with Trizol reagent according to standard protocols. RNA quantity and quality was determined by Nanodrop measurements (Nanodrop 1000, Thermo Fisher Scientific, Wilmington, DE). Quantity ranged between 150 and 400 ng/μl. All samples had a 260/280 ratio > 1.8.
Real-time PCR
1000 ng of total RNA were reverse transcribed into cDNA. For real-time PCR 10 ng were used for each amplification reaction (performed in triplicate). Real-time polymerase chain reaction (PCR) was performed as described [
21]. Expression levels of ABCB1 mRNA levels were determined using the Maxima SYBR Green/ROX qPCR Mastermix (Thermo Fisher Scientific) with β-actin serving as a reference gene. ABCB1 primer sequence [
22,
23]: ABCB1 sense: 5’-CCCATCATTGCAATAGCAGG-3’ and ABCB1 antisense: 5’-GTTCAAACTTCTGCTCCTGA-3’. β-actin primer sequence: β-actin sense: 5’-GGATGCAGAAGGAGATCACTG-3’ and β-actin antisense: 5’-CGATCCACACGGAGTACTTG-3’. For determination of EGFR mRNA expression Taqman assays using Taqman Maxima Probe/ROX qPCR Mastermix (Thermo Fisher Scientific) were performed. Taqman probes for EGFR (Hs01076078_m1) and GAPDH (Hs99999905_m1) were purchased from (Applied Biosystems, Waltham, MA). Quantification of ABCB1 and EGFR mRNA expression was calculated by the comparative Ct method using β-actin and GAPDH as reference genes, respectively. Experiments were performed twice delivering comparable results.
Migration assays
In wound healing assays, scratches were applied to confluent cultures in six-well plates (tissue culture treated, CytoOne® Starlab, Hamburg, Germany) using a pipette tip. Medium was renewed and the indicated treatments were added. Scratches were photographed after 0, 24 and 48 h and wound closure was calculated from the micrographs with T-Scratch (Computational Science & Engineering Laboratory, ETH Zurich, Switzerland). After a migration period of 24 h and 48 h, migrated cells on the bottom of the filter (cell culture insert for 24-well plates, 8.0 μm pore size, Falcon™ ThermoFisher Scientific) were counted (4 optical fields per well). Additionally, the bottom wells were further cultivated for 7 more days, fixed with methanol, stained with crystal violet and cell clones counted microscopically.
Statistical analysis
Statistical analysis was performed using GraphPad Prism 5.0 software. All data are expressed as mean ± S.D. Statistical significance of differences was analysed by using unpaired Student ́s t-test and one-or two-way ANOVA as appropriate followed by Bonferroni post-tests. A p-value < 0.05 was considered statistically significant. Throughout the study the following classification is used: *, p < 0.05; **, p < 0.01 ***, p < 0.001.
Discussion
In accordance with previous studies on human osteosarcoma cell lines and tissues [
6,
25], we found frequent and profound expression of EGFR in osteosarcoma cells suggesting that it might be an attractive therapy target. However, the actual functional contribution of EGFR to osteosarcoma cell biology has remained widely unexplored and literature regarding its prognostic value is contradictory. Some authors did not find a relation between EGFR expression and prognosis or treatment outcome [
6,
26]. In contrast, high EGFR expression was associated with significantly shorter survival times and disease free intervals in canine osteosarcomas [
27]. Furthermore, Wen et al. found increased EGFR levels in metastases and local recurrences when compared with primary tumors [
5]. Surprisingly, better prognosis was reported for patients with higher EGFR expression by Kersting et al. [
28] A shorter allele length of a CA repeating sequence in the intron I of the EGFR gene tended to be associated with increased EGFR expression. However there was no correlation between allele length and neoadjuvant chemotherapy response or long-term clinical outcome [
29]. Additionally, targeting of EGFR by cetuximab was suggested to induce osteosarcoma growth inhibition via antibody-dependent tumor cell phagocytosis by a M2-like macrophage subpopulation [
30].
In our study we could demonstrate functionality of the EGFR in osteosarcoma cell lines leading to EGF-mediated activation of the MAPK and PI3K/AKT pathways. Moreover, blockade of EGFR by gefitinib inhibited osteosarcoma cell proliferation at standard culture conditions only at comparably high concentrations not likely to be reached in vivo. This suggests that EGFR is not a central driver of osteosarcoma cell proliferation under in vitro cell culture conditions. Accordingly, gefitinib up to 20 μM did not reduce viability of several osteosarcoma cell lines in a previous study [
6]. However, under reduced serum concentrations, mimicking the starvation situation inside tumor nodules in vivo, the IC
50 values for gefitinib distinctly dropped. Also in other cellular stress conditions - including very sparse seeding in clonogenic assays and chemotherapy-induced cytotoxicity - EGFR-mediated signals supported osteosarcoma cell survival. At serum-reduced conditions, EGFR inhibition by gefitinib induced marked apoptotic osteosarcoma cell death. Additionally, gefitinib led to accumulation of cells in either G0/G1 or S phase of the cell cycle. Comparable observations have been published as a consequence of imatinib mesylate exposure of osteosarcoma cells. Interestingly, the authors suggested EGFR as one important target inhibited by this clinically approved kinase inhibitor [
31]. Moreover, osteosarcoma cell migration was significantly reduced by EGFR blockade and this effect was synergistically enhanced in combination with chemotherapy. These data are in accordance with a previous study demonstrating that selective in vitro inhibition of EGFR induced decreased motility, colony formation and invasiveness of osteosarcoma cells [
32]. The non-selective pan-erbB inhibitor CI-1033 achieved IC
50 values at concentrations of approximately 1 μM after 4 days of drug exposure [
33]. Pahl et al. showed cytotoxic effect of the anti-EGFR monoclonal antibody cetuximab induced by natural killer cell activation and following lysis of osteosarcoma cells [
34]. In both latter studies cytotoxic activity was not correlated with the level of EGFR expression. This is similar to our observations that neither the growth-inhibitory nor cell death-inducing effects of gefitinib were depending on the level of endogenous EGFR expression. Additionally, EGFR levels did not correlate with the histology of the original tumor. Interestingly exogenous EGF upregulated viability of osteosarcoma cells in the majority but not all cell lines tested. Thus, the highly EGFR-positive and gefitinib-sensitive OS-10 cell line was surprisingly not responsive to recombinant EGF under serum starvation. This suggests that EGFR might be stimulated even under serum-starved conditions by endogenous ligands in this cell model. Correspondingly, gefitinib reduced residual PI3K/AKT pathway activity determined as phosphorylation of AKT, S6 and GSK3β under serum-starved conditions. This strongly suggests that autocrine stabilisation of the PI3K/AKT pathway via EGFR supports starvation resistance of osteosarcoma cells as also observed in glioblastoma cells [
20]. Accordingly, activation of PI3K/AKT pathway and osteosarcoma cell migration was recently demonstrated by endogenous expression of transforming growth factor α via binding to EGFR [
35].
In many clinical treatment schemes, combination approaches of oncogenic kinase inhibitors and cytotoxic drugs turned out to be highly successful [
36‐
38]. Thus, we investigated combination of EGFR inhibition by gefitinib with osteosarcoma standard chemotherapeutics. We detected distinct synergism for combination with doxorubicin and MTX in osteosarcoma cells. ABCB1 efflux pump expression has been suggested as a major regulator of osteosarcoma chemotherapy response [
4]. As gefitinib is known to block ABCB1 efflux function [
39], the observed synergism with the ABCB1 substrate doxorubicin might be based rather on enhanced drug accumulation as blockade of an EGFR-mediated survival function. However, ABCB1 gene expression detected by real-time PCR was comparably low in all osteosarcoma cell lines making this assumption unlikely. Accordingly, synergism was also found for the none ABCB1 substrate MTX [
40], while the combination with cisplatin, also not transported by ABCB1, was widely antagonistic. Several previous preclinical studies on other tumor entities have shown additive and super-additive but also antagonistic interactions for EGFR inhibitors and chemotherapy [
41,
42]. Repeatedly, studies on EGFR targeting combined with cytotoxic drugs have been confirmed clinically, the most convincing being the therapeutic success achieved by the cetuximab-irinotecan combination in colorectal cancer [
43]. Combination of gefitinib with doxorubicin in soft tissue sarcomas showed synergistic impacts on cell proliferation and apoptosis induction compared to single agents. Furthermore, in animal models for soft tissue sarcoma combined low-dose doxorubicin and gefitinib were markedly synergistic [
44].
Two clinical phase I studies with advanced solid tumors revealed partial responses for a combination therapy of the EGFR inhibitors gefitinib and erlotinib with both cediranib (VEGFR inhibitor) and temozolomide [
45,
46]. Only 3 patients with osteosarcoma were included in these studies. One of them developed a partial response, which is encouraging for further studies on drug combination including EGFR tyrosine kinase inhibitors in osteosarcoma.
Competing interests
All authors declare that they have no competing interests.
Authors’ contributions
FS, LM were involved in performing the experiments, analysis and interpretation of the data, writing, reviewing, and revision of the manuscript. BK, RW, MM were involved in the design of the study and reviewing the manuscript. DL, SS, CP were involved in performing the experiments, analysis and interpretation of the data. WB was involvend in design of the study, analysis and interpretation of the data, writing, reviewing, and revision of the manuscript. All authors read and approved the final manuscript.