Background
Osteosarcoma (OS) is the most common primary malignant bone tumor. Because of a high rate of metastatic spread, curative treatment with surgery alone is rare. The long-term survival of patients with OS has improved during the last 40 years from 20% to nearly 80% due to the use of neoadjuvant chemotherapy in combination with surgery [
1]. However 20% of patients without evidence of metastasis at diagnosis either relapse locally or develop systemic disease preferentially by metastasis to the lungs or, less frequently, distant bones [
2]. This disease progression is primarily related to poor chemotherapy response resulting in a very unfavourable prognosis [
1,
3]. So far, histological response to neo-adjuvant chemotherapy is the most reliable predictor of survival in non-metastatic OS patients. Common chemotherapeutic regimens in OS include methotrexate, doxorubicin and cisplatin. However, if the tumor does not respond well to initial standard chemotherapy, no real alternative treatment option exists so far. Moreover, several previous studies demonstrated that a more aggressive and intensified chemotherapy could only improve histological tumor response but failed to improve patient survival [
4]. Therefore the implementation of new, preferentially bone-targeting drugs that can overcome chemotherapy resistance and inhibit metastasis especially to the lung are highly desirable and could further increase survival in OS patients [
5].
The antitumor activity of several gallium, germanium, titanium and ruthenium compounds has been recognized for some time, but none of these compounds have reached clinical routine so far [
6,
7]. In particular, simple gallium salts (gallium nitrate and gallium chloride) have been evaluated for their safety and efficacy in several clinical studies [
8]. Despite pronounced single-agent activity in lymphomas and bladder cancer and evidence for synergistic effects in combinations with established drugs, unfavorable pharmacokinetic and toxicological properties have prevented their routine use. Consequently, an organometallic gallium complex - KP46 (tris(8-quinolinolato)gallium(III)- with oral bioavailability was developed and has reached clinical evaluation [
9]. KP46 is 10-times more active than gallium chloride in vitro and has been reported to circumvent different forms of drug resistance [
10]. The great affinity of gallium and KP46 for bone [
11] clearly suggests that this gallium compounds might bear the potential to treat malignant bone tumors like OS. The effects of gallium salts on human cells are numerous and various [
8]. It has been widely accepted that a main antitumor activity is based on the inhibition of the iron-dependent enzyme ribonucleotide reductase. Due to its ability to compete iron uptake, gallium salts and also KP46 affects intracellular iron pools, but may also interact directly with ribonucleotide reductase displacing iron from the R2 subunit of this enzyme [
12,
13]. Additionally, gallium modifies the three-dimensional structure of DNA and blocks replication, modulates protein synthesis, and inhibits the activity of other enzymes, such as ATPases, DNA polymerases, and tyrosine-specific protein phosphatase. Furthermore, antimitotic effects of gallium have been described [
14]. Gallium complexes with lipophilic ligands - such as KP46 - have been developed to improve intestinal absorption without altering the pharmacodynamic effects [
11]. KP46 has already been successfully tested in a phase-I clinical trial on renal cell cancer and the evaluation of KP46 in a phase-II clinical trial was recommended [
15].
Proteins of the Bcl-2 family play an important role in regulation of programmed cell death and are possible future candidates for targeted therapy in OS. Recent studies demonstrated that gallium-induced cell death in lymphoma but also cytotoxic activity of KP46 against lung and colon cancer cells might involve pro-apoptotic Bcl-2 members like Bax and Bim triggering the mitochondrial apoptotic pathway [
16,
17]. Interestingly, increased expression of the anti-apoptotic Bcl-xL was associated with poorer survival of OS patients [
18] while Bcl-xL inhibition significantly enhanced chemo- and radiosensitivity of OS cells in vitro [
19]. Together these data suggest that inhibition of anti-apoptotic Bcl-2 members might be a feasible strategy to sensitize OS cells against chemotherapy.
Considering this information, aim of this study was to evaluate the possible antitumor effect of the bone-targeting gallium complex KP46 against human OS cells in vitro and to gain insight into the underlying molecular mode-of-action. Additionally, we set out to evaluate possible synergistic effects of KP46 with standard chemotherapeutics and compounds targeting apoptosis inhibition by members of the Bcl-2 family.
Methods
Chemicals
The organometallic gallium compound KP46 (tris(8-quinolinolato)gallium(III) was synthesized at the Institute of Inorganic Chemistry, University of Vienna (Vienna, AT). Doxorubicin, methotrexate (MTX), chloroquine, bafilomycin A1 and cisplatin were purchased from Sigma (St. Louis, MO), and obatoclax mesylate (GX15-070) from Selleck Chemicals Inc. (Houston, TX). Cisplatin was dissolved to a 5 mM stock in dimethylformamide (DMF, Sigma, St. Louis, MO), MTX to a 10 mM stock in NaOH and all other substances to 4–10 mM stocks in dimethylsulfoxide (DMSO, Sigma, St. Louis, MO). Stock solutions were diluted into culture medium immediately before use to obtain indicated concentrations. Final concentrations of solvents (DMSO, DMF, NaOH) were always less than 1% and tested for anti-OS activity in parallel.
Cell culture
OS cell lines MG-63, HOS, Saos-2 and U-2 OS were obtained from the American Type Culture Collection (ATCC, Manassas, VA). The human lung fibroblast (HLF) culture was established from a non-malignant lung surgery specimen. Cells were cultured in the respective growth media (Saos-2 in Mc Coy’s 5A medium, U-2 OS in Iscove’s Modified Dulbecco’s Medium and all other cell lines in RPMI1640 medium) supplemented with 10% fetal calf serum (FCS), obtained from PAA, Pasching, Austria, at 37 °C in 5% CO2 and regularly checked for Mycoplasma contamination.
Cell viability assay
Cells were seeded (2 × 10
4 cells/ml) in 100 μl growth media per well in 96-well plates. After a recovery period of 24 h, cells were treated with the indicated concentrations of the investigated drugs added to the cells in another 100 μl growth medium. If not indicated otherwise, drug exposure time was always 72 h. Cell viability was measured by the 3-(4, 5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT)-based vitality assay (EZ4U; Biomedica, Vienna, Austria) following the manufacturer’s recommendations. Cytotoxic effects were calculated with Graph Pad Prism software 5.0 (using a point-to-point function) (La Jolla, USA) and were expressed as IC
50 values calculated from full dose-response curves (drug concentrations inducing a 50% reduction of cell number in comparison to untreated control cells cultured in parallel). Values given are derived from at least three experiments performed in triplicates. Drug interactions in combination experiments were estimated using CalcuSyn software (Biosoft, Ferguson, MO) as described [
20,
21] and expressed by the combination index (CI) with CI < 0.9 representing synergism, CI 0.9–1.1 additive effects and CI > 1.1 antagonism.
Cells were plated (1 × 10
3 cells/ml) in 500 μl in 24-well plates and allowed to recover for 24 h. Drugs were added in 100 μl growth medium as indicated and cells were exposed to drugs for 7 days. After the drug exposure period, cells were washed with phosphate-buffered saline (PBS), fixed with methanol at 4 °C and stained with crystal violet. Clone area/μm
2 was determined using high-resolution pictures (Nikon7100) of at least 4 wells derived from two independent experiments in duplicate using Image J software. Moreover, single colonies >15 cells were counted using ImageJ Java software as described [
22]. Experiments were performed in duplicate and repeated twice.
Hoechst 33258/propidium iodide (HOE/PI) staining
OS cell lines were seeded (5 × 10
4 cells/well) in 24-well plates and exposed to KP46, obatoclax or a combination of both drugs at the indicated concentrations for 24 or 48 h exposure time. After the indicated incubation times, cells were stained with 2 μl/ml Hoechst 33258/propidium iodide mix (HOE/PI; HOE 1 mg/ml in PBS/PI 2.5 mg/ml in PBS), and incubated for at least an hour before microscopical evaluation using a Nikon Eclipse Ti inverted microscope (Vizitron Systems, Germany) [
22]. Positive staining with PI indicated dead cells (necrotic or late apoptotic). Nuclei of viable cells exhibited even blue fluorescence based on DNA staining by HOE. Bright blue fluorescence in condensed chromatin of PI-negative cells indicated mitosis characterized by regular mitotic features or early apoptosis based on small condensed nuclei or formation of apoptotic bodies. The number of viable undamaged, mitotic and apoptotic cells were counted in four optical fields per experimental group and well from two experiments in duplicate (ImageJ, Java Software) to quantify the photomicrographs.
Cell migration assays
Twenty-four-well plates were filled with 800 μl growth medium with 10% FCS, inserts (cell culture insert for 24-well plates, 8.0 μM pore size, Falcon™ ThermoFisher Scientific) were placed in wells and filled with 300 μl cell suspension (1 × 105 cells/ml) in growth medium without FCS. Drugs were added to the insert and well underneath. After 24 h drug exposure and migration time, inserts were removed. Lower wells were incubated with fresh culture medium for an additional 7 days, washed with PBS, fixed with methanol and stained with crystal violet. Wells and inserts were photographed and cell clones counted as mentioned above for the clonogenic assay.
Analysis of autophagy by Western blot
OS cells were seeded, proteins extracted after 24 h drug exposure and processed for Western blotting as described [
22]. Microtubule-associated protein 1 light chain 3 (LC3B), ATG5, ATG7 and Beclin-1 primary antibodies were purchased from Cell Signaling Technology (Danvers, USA). Monoclonal mouse antibody β-actin was obtained from Sigma. All primary antibodies were used as 1:1000 working dilutions and incubated overnight at 4 °C. Horseradish peroxidase-conjugated secondary antibodies [
23] (Santa Cruz Biotech, Dallas, USA) were used as 1:10.000 working dilutions. Autophagic flux was determined based on the method published by Chittaranjan et al. [
24]. In short, the lowest bafilomycin A1 dose inducing maximal LC3-II accumulation was determined by dose-response analyses and found to be 5 nM in both cell lines analysed (HOS, U-2 OS). Then KP46 at 1 and 10 μM was combined with bafilomycin (5 nM), obatoclax (250 nM) or both for 24 h and alterations in autophagic protein expression and LC3-I cleavage determined. Densitometric analysis of Western blot bands was performed using Image J software as published [
23] and protein expression quantified relative to β-actin in all cases. At least three Western blots from two independent experiments were analysed.
Acridine orange (AO) staining
Autophagosomes were stained with acridine orange (AO; 1 μg/ml), Merck, Darmstadt, Germany) after 24 h drug exposure on confluent cells (2 × 10
5 cells/ml). Stained cells were observed under the microscope and microphotographs taken (Nikon Eclipse Ti, Life-Cell Imaging) before dilution in FACS-PBS. Flow cytometry was performed using LSRFortessa flow cytometer and data analyzed with Flowing Software (University of Turku, Finland) [
25]. Fluorescence intensity was measured with the following filter settings: FL3 – red (488 nm/670 nm) and FL1 –green (488 nm/533 nm). The volume of the acidic compartment was estimated by the fluorescence intensity ratio FL3/FL1.
Cell cycle analysis
Cells (2 × 10
5 cells/well) were incubated for 24 h with indicated concentrations of KP46 or obatoclax in 6-well plates at 37 °C. After drug exposure, cells were collected and fixed in 70% ethanol at −20 °C for 1 h. RNase (0.2 mg/ml, Sigma, St. Louis, MO) and PI (0.01 mg/ml, Sigma, St. Louis, MO) diluted in FACS-PBS were used to degrade RNA and stain DNA, respectively. Cell cycle progression was examined by flow cytometry using FACS Calibur (Becton Dickinson, Palo Alto, CA) as described in Hoda et al. [
26]. CellQuest Pro software (Becton Dickinson) was used to analyze the resulting DNA histograms.
Statistical analysis
Data are presented as means ± SD of at least three experiments performed in triplicate, unless stated otherwise. Statistical significance between treatment groups was calculated with Graph Pad Prism 5.0 using t-test or one-way analysis of variance (ANOVA) including Bonferroni post-tests as appropriate. P-values below 0.05 were considered as statistically significant (* p < 0.05; ** p < 0.01; *** p < 0.001).
Discussion
In previous studies, KP46 demonstrated strong anti-cancer effects against cell lines and xenograft models of different human cancer types including colorectal cancer and melanoma [
17,
30,
31]. Furthermore, KP46 accumulates selectively in bone tissues [
11] suggesting an inherent tropism for primary and secondary bone tumors including OS. We therefore evaluated the activity of KP46 against a panel of human OS cell lines as single agent and its interaction with clinically used OS chemotherapeutics as well as experimental anticancer drugs. KP46 exerted profound cytotoxicity against all tested OS cell lines in the low micromolar to nanomolar range depending on the assay format and exposure time. Besides induction of cell detachment and death at higher concentrations, treatment with low-dose KP46 additionally reduced OS cell migration and clonogenic survival. In contrast, changes in cell cycle distribution during KP46 treatment were comparably minor and consisted of an accumulation in S-phase as has also been reported in previous studies in other cancer types [
30,
31]. Despite a significant difference in the sensitivity of the tested OS cell models against the cytotoxic effect of KP46, all tumor cell lines were distinctly hypersensitive as compared to the non-malignant HLF cell line. This may indicate a specific tumor-targeting effect of KP46 and suggests existence of a therapeutic window for this bone-targeting compound at least at the cellular level. Interestingly, cisplatin exhibited the lowest IC
50 values in the non-malignant HLF cell model well in agreement with the massive adverse effects by this clinically used anticancer metal drug.
The molecular mechanisms underlying the anticancer activity of KP46 are still not well understood. Some studies - including several from our group - suggested apoptotic cell death induction as one major mode of action for KP46 especially after longer drug exposure times [
13,
30,
31]. A recent in vitro study on colon cancer cells has proposed Ca
2+-release-mediated p53-dependent and –independent pathways of KP46-mediated programmed cell death induction. While in p53
+/+ cells p53-induced upregulation of reactive oxygen species (ROS) was proposed as central mode-of action downstream of intracellular Ca
2+-release, in p53-mutated or -deleted cells FAS-related extrinsic apoptosis was postulated [
31]. Accordingly, a p53-dependent cell death mechanism via KP46 accumulation in mitochondria and deregulation of mitochondrial dynamics and bioenergetics was suggested recently for human colon cancer cells [
13]. Depletion of the intracellular labile iron pool by KP46 initiated p53-dependent BNIP3L activation and in turn mitophagic cell death [
13]. However, in a previous study we were unable to detect any evidence for an impact of the p53 status on the general sensitivity of an extended panel of lung and colon cancer cells against KP46 treatment [
17]. Accordingly, of the here investigated OS cell lines only U-2 OS harbors an intact p53 response while the other three cell models are p53 mutated or null [
32,
33]. However, U-2 OS was characterized by the lowest sensitivity against KP46 arguing against a major sensitizing role of wild-type p53 against this gallium-containing metal compound. Consequently, though p53-mediated stress or damage recognition might impact on the mode of KP46-induced cell death, it is obviously not a general determinant of cancer cell sensitivity to KP46.
OS cells responded to KP46 by a rapid loss of cell adhesion and rounding up with a subpopulation of dead cells lacking classical features of apoptosis like chromatin condensation and formation of apoptotic bodies. Recently, our group has described a comparable anoikis-like cell death in KP46-hypersensitive colon and lung cancer cell models [
17]. This additional form of KP46-induced cell death was insensitive to caspase-inhibitors but characterized by distinct downregulation of integrin β1 accompanied by cleavage of talin in a calpain inhibitor-sensitive fashion. Calpain regulates integrin turnover through degradation and cleavage of talin, the binding partner of integrin [
34,
35]. Whether loss of integrin β1 expression, also detected in KP46-treated OS cells (data not shown), is the underlying mechanism is matter of ongoing investigations. Additionally to the rapid loss of cell adhesion, already low, subtoxic KP46 concentrations significantly blocked OS cell migration. As metastatic spread is still the leading cause of death for OS patients [
2], these anti-adhesive and anti-migratory effects of KP46 might be especially interesting to develop this bone-targeting gallium compounds for systemic treatment of OS.
Combined chemotherapy has dramatically improved outcome of OS patients, however, relapse occurs is a certain percentage of cases based on resistance development. Hence, therapy response is a major prognostic factor for OS patient survival [
27]. Consequently, we tested whether KP46 might represent a feasible candidate for anti-OS combination strategies. In our setting, synergistic effects were most pronounced when KP46 was used together with the standard OS agent cisplatin. Comparable observations have been reported in early studies for the platinum drugs cisplatin, carboplatin and oxaliplatin in ovarian and colon cancer cells [
9]. The underlying molecular mechanisms of this synergism are currently enigmatic and might include interactions with repair mechanism or programmed cell death regulatory signals. Accordingly, we found upregulation of pro-apoptotic Bcl-2 family members Bax and Bim as well as enhanced Bax mitochondrial translocation in KP46-treated colon and lung cancer cells [
17]. Pro- and anti-apoptotic Bcl-2 family members and especially expression of Bcl-2 and Mcl-1 are also decisive in the regulation of cisplatin sensitivity of OS cells [
36,
37]. Consequently, we also combined two inhibitors of anti-apoptotic Bcl-2 family members, namely the BH3 mimetics obatoclax and venetoclax (ABT-199), with KP46. Unexpectedly, only the former drug, being a relatively broad Bcl-2 inhibitor, strongly synergized in the majority of OS cells with KP46, while the relatively Bcl-2-specific venetoclax failed to enhance the anti-OS activity. This argues against a major inhibitory function of anti-apoptotic Bcl-2 family members on KP46-induced OS cell death.
Additionally to Bcl-2 inhibition, obatoclax but not venetoclax is known to effectively block autophagy by interfering with autophagosomal acidification [
28]. Especially in the less KP46-sensitive OS cell model U-2OS, KP46 clearly induced autophagy by appearance of larger acidic vesicles and accumulation of the processed form of LC3B, the phosphatidylethanolamine-bound LC3-II, a widely used autophagy marker. Combination experiments suggested an enhanced autophagic flux by KP46 which could be blocked by combination with downstream autophagy inhibitors obatoclax and chloroquine. Indeed, KP46-induced macroautophagy and mitophagy were also recently described in HCT116 cells, the latter representing a mode-of-action contributing to the cytotoxic activity [
13]. Accordingly, mitophagy inhibition by downregulation of BNIP3L rescued cells from rapid KP46-mediated cell death. However, on the contrary, macroautophagy is a well-described resistance mechanism also of OS cells against a variety of synthetic and natural anticancer drugs [
38,
39] suggesting that autophagy inhibition might synergize with KP46. Furthermore, there is evidence that the inhibition of integrin function and thus loss of cell adhesion, leads to protective autophagy induction and enables survival if cell re-adhesion is possible. Additionally, autophagy seems to be an important component in the life cycle of integrins and has a direct impact on cell migration [
38,
40‐
44]. Therefore we hypothesized that autophagy inhibitors should sensitize OS cells to the anti-tumor activity of KP46. Indeed, not only obatoclax but also the classical late stage autophagy inhibitor chloroquine demonstrated comparable synergistic effects in the investigated OS cell lines with the exception of SAOS-2 cell, where the effects were additive. Together this strongly suggests that the inhibition of autophagic flux by obatoclax and not its interaction with bcl-2 family members is underlying the synergism with KP46. Consequently, at least in OS cells macro-autophagy is obviously a way to remove damages induced by KP46 to allow cancer cell survival.
Acknowledgements
We thank Mirjana Stojanovic for technical assistance.