Drug Repurposing of β-blocker Propranolol in Osteosarcoma: Preclinical Antitumor Efficacy Alone or in Combination with Chemotherapy.


 Background: Osteosarcoma is still associated with poor response to therapy and alarmingly elevated mortality rates, especially in developing countries, highlighting the urgent need of novel therapeutic strategies. Given that β-adrenergic receptor (β-AR) signaling regulates many cellular processes involved in the initiation and progression of cancer, multiple efforts have been made to repurpose non-selective β-blocker propranolol in oncology. Considering the unsatisfied clinical needs of osteosarcoma, we evaluated the in vitro and in vivo antitumoral activity of propranolol as a monotherapy or in combination with metronomic chemotherapy, in β-AR-expressing human osteosarcoma cells. Methods: In vitro anti-osteosarcoma effects of propranolol were assessed on high-density cell growth, chemotaxis, colony-forming ability and three-dimensional spheroids. Cell cycle phase distribution and apoptosis were evaluated using flow cytometry analysis, and quantification of hypodiploid cell population and TUNEL labelling, respectively. Propranolol addition to chemotherapy was evaluated in vitro and in human osteosarcoma xenografts generated in athymic mice.Results: Propranolol significantly impaired MG-63 and U-2 OS cellular growth in a concentration range of 10-100 µM (IC50 ≈ 45µM) and was capable of blocking in vitro protumoral effects triggered by catecholamines. Cytostatic activity of propranolol was associated to reduced mitosis and G0/G1 cell cycle, but not to apoptosis induction. Moreover, β-blocker drastically reduced colony formation, spheroid progression and migration of osteosarcoma cells. Synergistic effects were observed in vitro after combining propranolol with chemotherapeutic agents methotrexate or cisplatin. In vivo, treatment with propranolol (10 mg/kg i.p.), alone and especially in addition to non-interrupted low-dose cisplatin (2 mg/kg i.p.), markedly abrogated xenograft progression, reducing tumor growth rates by 25 or 70%, respectively. After histological analysis, propranolol and cisplatin combined therapy resulted in low tumor mitotic index and increased tumor necrosis. Conclusions: Considering its robust antitumoral effects, β-AR antagonization using β-blocker propranolol seems to be an achievable and cost-effective therapeutic approach to modulate osteosarcoma aggressiveness, in addition to chemotherapy. Propranolol could be proposed as a co-adjuvant agent for the management of osteosarcoma, especially in low- and middle-income countries where access to treatment resources are limited. Further preclinical and clinical studies of propranolol repurposing in osteosarcoma are warranted.


Background
Osteosarcoma (OSA) is the most prevalent primary bone tumor, affecting mostly pediatric and juvenile age-group patients [1]. Despite global 5-year overall survival in OSA patients is near to 70%, there is a welldocumented disparity regarding patient survival between high-income countries (HIC) and low-middle income countries (LMICs) [2]. In addition to the surgical removal of primary and secondary lesions, management of OSA involves adjuvant administration of chemotherapy, including cisplatin and methotrexate [3]. However, chemotherapy effectiveness in OSA patients is restricted by the development of drug resistance by tumor cells, lack of selectivity and high toxicity [4,5]. Taking this into account, there is an urgent need to develop novel, cost-effective and selective therapeutic alternatives for the treatment of aggressive OSA, putting emphasis on relevant biological events that rule disease progression.
Drug repurposing in oncology is de ned as the implementation of already-approved drugs for other medical applications to treat cancer. In comparison to de novo drug discovery, both development risks and costs, as well as safety-related failure, are reduced with this approach because such drugs have wellknown pharmacokinetic/pharmacodynamics pro les, and pharmacovigilance for adverse effects is largely accessible [6,7]. Moreover, seeking for enhanced therapeutic bene ts, repurposed drugs are often combined with metronomic chemotherapy, consisting in frequent administrations of low-dose chemotherapy, with no sustained drug-free breaks [8]. This treatment strategy is called "metronomics", and stands as an interesting therapeutic approach to especially implement in LMICs as its reduced costs and fewer infrastructure needs can be easily adapted in resource-limited countries [9].
Propranolol (PPN) is a non-selective β-adrenergic receptor (β-AR) antagonist that was originally used for the treatment of hypertension and other cardiovascular conditions [10]. Over the last years, PPN has demonstrated potent anticancer effects in a wide variety of oncological indications. After initial clinical retrospective observations and subsequent clinical trials, this β-blocker was rst repurposed for the treatment of infantile hemangioma, a common vascular tumor of infancy [11]. Preclinical studies showed that β-AR type 1 and 2 blockade using PPN results in cell cycle arrest, apoptosis and proliferation reduction of breast, melanoma and liver tumor cells, among others, mainly by disrupting mitogenic signaling cascades involving protein kinase B (PKB, AKT), mitogen activated protein kinase (MAPK) and cyclic adenosine monophosphate (cAMP) [12][13][14]. Moreover, PPN demonstrated to reverse protumoral effects mediated by stress hormones in chronic or surgical stress models, reducing angiogenesis and metastasis, and increasing immune function [10,15]. Taking all these data into account, assessment of PPN anticancer properties has recently reached clinical stages for angiosarcoma, ovarian cancer and melanoma, mainly as a surgical adjuvant or in combination with chemotherapy or biologic targeted therapy [16-18]. Despite sympathetic nervous system (SNS) activation and signaling pathways triggered by catecholamines such as epinephrine and norepinephrine are deeply involved in cancer initiation and progression [19,20], little is known about the speci c role of catecholamines or β-AR signaling in OSA. As recently reported, expression of β-ARs and several catecholamines, including norepinephrine, was signi cantly higher in OSA in comparison to healthy bone [21]. In addition, all subtypes of β-AR were found to be expressed in a wide collection of human and canine OSA or other bone sarcoma models [22][23][24], making the exploration of non-selective β-blocker PPN in OSA an appealing niche for further studies. Considering the unsatis ed clinical needs of OSA and the multiple reported bene ts of PPN as a broad antitumor agent, the aim of the present study was to evaluate the in vitro/in vivo antitumoral activity of PPN in OSA, as a monotherapy or in combination with metronomic chemotherapy.

Methods
Drugs and compounds. DL-PPN hydrochloride lyophilized powder (Sigma-Aldrich, Missouri, US) was rst solubilized using citrate buffer (pH 3) and further diluted with phosphate buffer saline (PBS) to reach working concentrations. Cisplatin (CDDP) and methotrexate (MTX) were purchased from Fada (Buenos Aires, Argentina) and Abbot laboratories (Illinois, US), respectively. Catecholamines epinephrine and norepinephrine were purchased from BIOL laboratories (Buenos Aires, Argentina).
All cells were harvested using a trypsin/EDTA solution (Gibco, Massachusetts, US) and routinely tested por mycoplasma.
OSA cell growth and citotoxicity. Direct in vitro effects of PPN, alone or in addition to chemotherapy, or catecholamines, against rapidly growing MG-63 or U-2 OS OSA cells were assessed by crystal violet staining after a 72-h treatment (for detailed methodology see Additional le 4. Supplementary methods). Before evaluating the effects of PPN on cell cycle phase distribution or migration, lack of direct short-term cytotoxicity of PPN was con rmed using the trypan blue dye exclusion assay (for detailed methodology see Additional le 1. Supplementary methods).
Cell cycle distribution. Cell cycle phase distribution and hypodiploid population quanti cation in starved OSA cells following a 24 h PPN (50 µM) treatment was evaluated by ow cytometry as previously In vitro mitosis. PFA-xed and DAPI-stained OSA cell cultures were analyzed for mitotic index calculation after a 48-h exposure to PPN (50 µM). Mitotic gures in all four basic stages of mitosis were counted using a high-power eld (HPF) of 400-fold magni cation (X400) and the percentage of mitotic cells was obtained after quanti cation of total cell number. Ten pictures per group were used for quanti cation.
Clonogenic growth. Effect of PPN (1, 10 or 50 µM) on OSA colony forming ability was assessed using a 7-d clonogenic assay, as previously reported [25] (for detailed methodology see Additional le 1.

Supplementary methods).
OSA spheroid growth. Impact of PPN (50 µM) treatment on 3D growth was evaluated on MG-63 multicellular spheroids generated by the hanging drop method [26]. After con rming appropriate sphericity and size (> 200 µm), fully-formed spheroids were incubated with PPN during one week, and volume was periodically assessed as a direct parameter of 3D OSA growth (for detailed methodology see Additional le 1. Supplementary methods). Animals. Outbred athymic female N:NIH(S)-nu mice aged 8 weeks with a weight of approximately 23 g, were purchased from the School of Veterinary Sciences Animal Facility at National University of La Plata (Buenos Aires, Argentina), and, after randomization, housed at 5 mice per cage in our animal facility at the National University of Quilmes. Food and water were provided ad libitum and general health status of the animals was monitored daily.
OSA xenograft progression. Human OSA tumors were heterotopically generated after subcutaneous injection of MG-63 cells in athymic mice [28]. In vivo combination studies were performed by administering PPN 10 mg/kg i.p. in a 5-day-on, 2-day-off schedule, alone or in combination with 2 mg/kg i.p. doses of CDDP three times per week during 4 weeks, until the end of the protocol. PPN [29,30] and CDDP [31,32] dosage were de ned according to previously reported preclinical studies following a metronomic rationale. Tumor growth rates and volume, as well as total animal weight were recorded or calculated throughout the protocol. Histopathological assessment of OSA tumors involved mitotic index quanti cation in viable sections of H&E-stained tumor slides and determination of adjusted tumor necrotic rate after treatment [33] (for detailed methodology see Additional le 1. Supplementary methods).
Statistics. Statistical analysis was performed using the PRISM 6 (GraphPad Software Inc., San Diego, US) or Compusyn software (Combosyn Inc., New Jersey, US). To compare differences between two experimental groups Mann Whitney or t tests were used for non-parametric or normal distribution of data, respectively. In case of more than two experimental groups, ANOVA analysis with Tukey's multiple comparisons post-test was used when normal distribution of data was determined. Kruskal-Wallis analysis with Dunn's multiple comparisons post-test was used in case of non-parametric distribution of data. Differences were considered statistically signi cant at a level of p < 0.05. Data correspond of at least 2 or 3 independent experiments unless stated otherwise. Data were presented as mean ± standard deviation (SD) or standard error of mean (SEM).

Results
First, in order to evaluate in vitro OSA sensitivity to PPN, semicon uent monolayers of rapidly growing MG-63 and U-2 OS cells were exposed to tested compound using a concentration range between 1 µM and 1 mM, as determined based on previously reported preclinical studies [13,14,34]. After a 72-h exposure to PPN, we observed a clear dose-dependent cell growth inhibition in both tested cell lines obtaining IC 50 values of 45.6 and 47 µM, respectively (Fig. 1a). At high concentrations however, MG-63 showed an increased sensitivity to PPN, obtaining more than a 95% growth reduction at 100 µM or higher. As expected, human glioma U-87MG cells, which do not express β1-AR or β2-AR [35], were exposed to 50 µM PPN (≈ IC 50 in responsive cells) and showed a complete lack of cytostatic effect by the compound (Fig. 1b). Both OSA cell lines are positive for β1-AR and β2-AR expression, and are responsive to catecholamine agonistic stimulation [22,23]. After 72-h stimulation, catecholamines epinephrine and norepinephrine increased OSA cell proliferation, showing a maximum protumoral effect at 10 nM (data not shown). In this context, coincubation with PPN using a low 10 µM concentration was capable of completely blocking the protumoral activity of catecholamines (Fig. 1c).
A signi cant arrest in the G 0 /G 1 cell cycle phase was observed following a 24-h treatment using a 50 µM concentration of PPN as evaluated by ow cytometry studies. This arrest was accompanied by a decline of cell populations in both S and G 2 /M phases (Fig. 1d). To discard any short-term direct cytotoxicity of PPN on such treatment schedule, semicon uent MG-63 cultures exposed to PPN were assessed by the trypan blue dye exclusion assay, showing that PPN at ≈ IC 20 and IC 50 concentrations had no effect on cell viability (data not shown). Interference of cell cycle progression after β-blockage was further con rmed by in vitro mitotic index calculation, in which PPN caused a dramatic 80% reduction in the % of mitotic cells in OSA cultures ( Fig. 1e and f). Despite the large amount of preclinical/clinical data reporting the pro-apoptotic effects of PPN on malignant cells [12,14], PPN at a concentration ≈ IC 50 wasn´t capable of inducing apoptosis in OSA cells, as evaluated by quanti cation of hypodiploid sub-G 0 /G 1 cell population and TUNEL labeling ( Fig. 1g and h).
Early metastatic spread and recurrences after tumor removal are strongly related to poor prognosis in OSA [36]. Considering that tumor cell migration, colony establishment and microtumor outgrowth are all hallmarks of the early stages of metastatic progression, we tested the effect of β-blocker PPN on the establishment and progression of 2D tumor colonies, 3D spheroid growth and OSA cell chemotaxis. Clonogenic growth of OSA cells was severely suppressed after 7 days of incubation with 1, 10 and 50 µM of PPN, reducing colony formation capacity by 31, 46 and 100%, respectively, in comparison to saline vehicle-treated cells (Fig. 2a). In this experimental setting, an IC 50 value of 19.26 µM was obtained on long-term cell cultures, a concentration 2.5-fold lower than that obtained for exponentially-growing cells. As above-mentioned, anti-OSA activity was also evaluated on established multicellular spheroids during a 7-day treatment period. In contrast to control spheroids which grew until reaching a 6-fold increase in volume, growth of PPN-treated spheroids was completely blocked (Fig. 2bi, ii and iii), as observed in representative photographs or individual spheroid progression curves. Chemotaxis of MG-63 cells after an overnight exposure to PPN was also signi cantly impaired, causing a cell migration reduction of 48% (Fig. 2c).
Anti-metabolite MTX and DNA-damaging agent CDDP remain as cornerstones in the treatment of OSA [3]. With the aim of evaluating potential therapeutic bene ts after addition of PPN to metronomic chemotherapy, we rst conducted in vitro combination studies exploring the cytostatic effects of PPN plus MTX or CDDP. After calculating IC 50 values for MTX and CDDP as monotherapies (Additional le 2. Fig. S1a and b), sub-and supra-IC 50 concentrations of PPN and cytotoxic agents were de ned for combination assays. Synergy was observed in all low-and high-concentration combination scenarios after addition of PPN to MTX (Fig. 3a) or CDDP (Fig. 3b) (Speci c CI and DRI values can be found in Additional les 3 and 4. Tables S1 and S2, respectively). Only in the PPN 50 µM plus MTX 10 nM combination, addition of PPN was unable to improve MTX anti-OSA activity (Fig. 3a, right). Taking these results into consideration, in addition to previous studies reporting cooperative activity between PPN and CDDP-based chemoradiotherapy in different tumor types [37,38], the anti-OSA activity of PPN in combination with CDDP was further evaluated in vivo on MG-63 xenografts (Fig. 4a). Treatment with PPN in addition to metronomic CDDP dramatically hindered OSA xenograft progression, reducing tumor growth rate by 70% in comparison to vehicle-treated mice and signi cantly improving the anti-OSA activity of PPN or CDDP alone (Fig. 4b). PPN treatment as a single agent was also capable of reducing tumor growth, causing a less profound reduction of about 20%. Therapeutic bene ts of PPN plus CDDP were con rmed after measuring nal tumor burden, were mean tumor weight was reported as 146.7 ± 38.9 mg in control group in comparison to 53 ± 23.8 mg (mean ± SD) in the PPN plus CDDP group (Fig. 4c). Furthermore, all treatments were well tolerated as no signi cant changes in animal body weight, food or water consumption were observed throughout the protocol (Fig. 4d). Representative OSA tumorbearing mice from different experimental groups are depicted in Fig. 4e.
Given that high mitotic counts are associated to OSA aggressiveness and poor prognosis [39], mitotic index was assessed in resected tumors obtaining values of 8.3 ± 2.7, 5 ± 2.3, 3.36 ± 2.1 and 2.81 ± 2.2 mitotic bodies/HPF for PPN, CDDP and PPN + CDDP groups, respectively ( Fig. 5a and b). Despite sustained PPN + CDDP therapy was associated to the most drastic drop in mitotic counts (67% inhibition), all tested therapies achieved signi cant reductions. Interestingly, despite not being re ected on its effect on OSA growth, sustained low-dose CDDP treatment was linked to an evident reduction in the proliferating status of OSA cells within the tumor, quite equivalent to combined therapy. Considering that assessment of tumor necrosis in OSA has become a relevant tool for evaluating response to therapy [40,41], adjusted tumor necrotic rates (ATNR) were calculated for all experimental groups (Fig. 5c). Only PPN plus CDDP-based therapy was capable of signi cantly enhancing ATNR, from 58.32 ± 5.04 to 81.74 ± 2.25%. Besides a marked increase in ATNR, tumors belonging to PPN + CDDP-treated animals displayed a peculiar diffuse and rather centered pattern of viable tumor tissue in contrast to control, PPN or CDDP groups, where viable tumor tissue areas were found as large and scattered basophilic regions, interrupted by vast areas of necrotic tissue (Fig. 5d).

Discussion
To the best of our knowledge this is the rst study to show therapeutic bene ts of non-selective β-AR antagonist PPN in combination with chemotherapeutic agents in human OSA models. Exposure of OSA cells to PPN was associated to reduced proliferation and mitosis, G 0 /G 1 cell cycle arrest and impairment of 2D/3D tumor colony formation and growth. With an IC 50 ≈ 50 µM, sensitivity of OSA cells to PPN appears equivalent or even higher compared to other aggressive pediatric solid tumor cell lines with reported IC 50 ranging from 114 to > 200 µM [14,34]. Despite no signi cant proapoptotic activity of PPN on OSA cells was observed at 50 µM, it is worth noting that apoptosis induction in other preclinical studies using PPN on non-OSA tumor models was achieved using concentrations in the range of 100-300 µM [12,14,34]. Exploring such concentrations goes beyond the scope of the present work given that, according to early pharmacokinetics reports, clinically relevant concentrations of PPN are in the low µM range [42,43]. PPN cytostatic activity in β-AR expressing malignant cells was previously associated to the down-regulation of the AKT/mammalian target of rapamycin (mTOR) and the MEK/ERK1/2 MAPK signaling pathways [12][13][14]. Interestingly, with the aim of shedding some light into the largely unknown molecular landscape of OSA, genomic and transcriptomic studies have recently identi ed frequently mutated genes and copy number alterations in OSA samples [44,45]. Besides con rming TP53, RB1 and MYC as driver genes, it was reported that PI3K/AKT/mTOR and MAPK pathways are heavily altered in OSA and are central players in disease initiation and progression. According to these data, PPN could be capable of blocking key protumoral signaling pathways in OSA biology. Thus, further studies exploring speci c molecular mechanisms associated to PPN anti-OSA effects are mandatory.
After addition of PPN to CDDP or MTX, synergistic anti-OSA effects were observed in all in vitro combination studies. Increased antitumor activity as a result of combining PPN with cytotoxic agents was previously reported in non-OSA preclinical and clinical studies, in which addition of repurposed drug to paclitaxel-, 5-FU-or even CDDP-based chemotherapy increased antiangiogenic and antiproliferative effects of monotherapies, as well as decreased distant metastases in patients [30,37,38]. Moreover, it was recently reported that β-blockade in different bone sarcoma models, increases sensitivity to radiotherapy, especially in canine OSA [24]. In our in vivo protocol sustained treatment with PPN, alone and especially in addition to CDDP, caused a marked decrease in OSA tumor progression, showing that PPN may complement chemotherapy, increasing its effectiveness without overt toxicity. After histological assessment of OSA tumors, we observed a signi cant increase in combined treatment-associated necrosis as well as a profound reduction in mitotic indices. These ndings bear high translational relevance given that necrosis rate in response to therapy and mitotic index determination are two histopathological parameters routinely used in the clinical management of OSA, linked to disease aggressiveness and prognosis [33,39]. Besides complementing chemotherapy, β-blockade using PPN could also show bene t during resection of primary or secondary OSA tumors in order to antagonize several deleterious mechanisms triggered by surgery, including in ammation and release of catecholamines, activation of proangiogenic and protumoral signaling pathways, tumor cell shedding, among others [46].
In our study, PPN anti-OSA activity was related to direct cytostatic effects and cell cycle arrest, decreasing survival and mitosis in β-AR expressing tumor tissue. However, SNS activation and β-adrenergic signaling can regulate a wide range of cancer-associated molecular pathways via both direct effects on β-AR expressing transformed cells and regulation of other β-receptor-bearing cells populating the complex tumor stroma, such as vascular or immune cells [47,48]. In spite of using T cell-de cient nude mice for our in vivo studies, this animal model still has a robust innate immune response involving dendritic cells, tumor-associated macrophages and natural killer cells, which are known to play a key role against OSA [49]. Taking these concepts into account, other microenvironmental mechanisms associated to pleiotropic β-blockade, such as angiogenesis inhibition or increased immune function, could be playing a part in the overall PPN anti-OSA activity and should not be discarded. For instance, PPN was shown to be highly effective and safe in the management of angiogenic diseases such as infantile hemangioma, halting the growth and ulceration of these vascular tumors [11,50].
Our study has several limitations worth noting. It is widely recognized that in vivo models should recapitulate the tumor stroma as closer to the clinical setting as possible [51]. Given that PPN targets both tumor and its microenvironment, the in uence of OSA tumor stroma should not be neglected, and the use of orthotopic paratibial or intraosseous tumor models should be considered for future studies.
Despite showing that β-blockade by PPN affects metastasis-associated cellular events in vitro, such as 2D/3D colony growth or tumor cell chemotaxis, effects of PPN administration on in vivo metastatic spread and growth, especially to lungs, should be evaluated in order to fully characterize compound´s therapeutic potential in OSA.
OSA is fatal for about one-third of the patients worldwide. However, prognosis is signi cantly poorer for LMICs [9]. Higher mortality rates are mainly due to economic inequalities and ine cient health-care systems, resulting in diagnosis at advanced stages and poor access to treatment resources, including surgery, radiotherapy and high-cost novel cancer drugs [2]. As previously introduced, metronomics combines drug repurposing with metronomic chemotherapy and could act as a substitute for standard treatments when unavailable or undoable [9]. PPN is a low-cost widely-available FDA-approved drug and its repositioned use in oncology is gaining strength as high quality preclinical and clinical evidence accumulates. On the other hand, in the past years metronomic chemotherapy has emerged as a promising multi-targeted therapeutic approach, showing clinical bene ts in aggressive pediatric tumors with low associated-toxicities and used in outpatient settings [8,[52][53][54][55]. Therefore, due to its costeffectiveness, good tolerability and accessibility, the combination of metronomic chemotherapy in addition to repurposed PPN could represent an interesting alternative strategy to manage OSA, especially in developing countries.

Conclusions
Survival rates of OSA patients have remained practically unaltered in the last 30 years, highlighting the need of intensifying research efforts and drug development programs to tackle this disease. β-AR antagonization seems an achievable and interesting therapeutic approach to inhibit OSA progression (Fig. 6). Integration of novel therapeutic approaches such as the use of repurposed PPN into current OSA management is both promising and challenging. We believe that this study could help lay the groundwork for translating the use of PPN as a novel adjuvant in OSA.
All animal protocols have been carried out in accordance with the Guide for the Care and Use of Laboratory Animals as adopted by the U.S. National Institutes of Health (NIH Publications No. 8023, revised 1978) and were approved by our institutional Animal Care Committee UNQUI-CICUAL (Resolution CD CyT Nº075/14); Protocol code: 011-15. Results from animal studies were reported following ARRIVE guidelines.
Consent for publication.

Not applicable
Availability of data and materials.
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Competing interests.
All authors have no con ict of interests to declare.   Synergistic anti-osteosarcoma activity of propranolol in combination with standard cytotoxic agents.
Tumor cell growth was evaluated on log-phase growing MG-63 cells and measured by crystal violet staining after a 72-h drug exposure. Combinational studies were conducted after adding PPN to (a) methotrexate (MTX) or (b) cisplatin (CDDP). To test the effect on OSA cell survival, drugs were combined using sub-IC50 (left) or supra-IC50 (right) concentrations. Data are presented as mean ± SEM and are representative of at least two or three independent experiments. *p<0.05, ***p<0.001, ****p<0.0001. ANOVA followed by Tukey´s test.  ANOVA followed by Tukey´s test. propranolol blocks protumoral effects of catecholamines, and inhibits osteosarcoma growth, especially