Contribution of Mitochondrial Activity to Doxorubicin-Resistance in Osteosarcoma Cells

Simple Summary Osteosarcoma represents the most common bone tumor, and it is the second most fatal cancer in children and young adults. However, despite the clinical benefit of chemotherapy, osteosarcoma patients still have a poor prognosis due to the common onset of drug resistance, as well as the disease’s growth and metastasis frequency. Thus, understanding the processes underlying the phenomenon of resistance represents a challenge in the field of oncology and may lead to new approaches to overcome resistance. This study highlights a different mitochondrial phenotype in doxorubicin-resistant cancer cells and suggests targeting the altered pathway to re-sensitize the resistant clones to the drug. Taken together, these results contribute to identifying the mitochondrial metabolism changes and to developing combined approaches with the final goal of overcoming doxorubicin resistance in osteosarcoma. Abstract Osteosarcoma is considered the most common bone tumor affecting children and young adults. The standard of care is chemotherapy; however, the onset of drug resistance still jeopardizes osteosarcoma patients, thus making it necessary to conduct a thorough investigation of the possible mechanisms behind this phenomenon. In the last decades, metabolic rewiring of cancer cells has been proposed as a cause of chemotherapy resistance. Our aim was to compare the mitochondrial phenotype of sensitive osteosarcoma cells (HOS and MG-63) versus their clones when continuously exposed to doxorubicin (resistant cells) and identify alterations exploitable for pharmacological approaches to overcome chemotherapy resistance. Compared with sensitive cells, doxorubicin-resistant clones showed sustained viability with less oxygen-dependent metabolisms, and significantly reduced mitochondrial membrane potential, mitochondrial mass, and ROS production. In addition, we found reduced expression of TFAM gene generally associated with mitochondrial biogenesis. Finally, combined treatment of resistant osteosarcoma cells with doxorubicin and quercetin, a known inducer of mitochondrial biogenesis, re-sensitizes the doxorubicin effect in resistant cells. Despite further investigations being needed, these results pave the way for the use of mitochondrial inducers as a promising strategy to re-sensitize doxorubicin cytotoxicity in patients who do not respond to therapy or reduce doxorubicin side effects.


Introduction
Osteosarcoma is a severe bone tumor mostly affecting children and adolescents [1]. Its worldwide incidence rate reaches four cases per million per year [2,3]. Current therapeutic management includes surgery, radiotherapy, and chemotherapy [2]. The standard of care for chemotherapy is represented by methotrexate, cisplatin, and doxorubicin, or a combination of these three agents [4].
Doxorubicin (DXR) is a chemotherapeutic agent intercalating DNA and belongs to the anthracycline family. The mechanism of action consists of the inhibition of DNA topoisomerase II, which blocks DNA replication hampering the recombination of DNA double-strand. Recently, another mode of action for DXR has been proposed: this mechanism involves increased reactive species of oxygen (ROS) levels which causes DNA damage and leads to cell death [5]. The successful use of doxorubicin in clinical therapy has been jeopardized by severe toxicity, such as cardiotoxicity [6], and the onset of drug resistance [7].
The main known mechanisms of DXR resistance involve increased drug efflux due to P-glycoprotein expression, which is also associated with a worse outcome [8,9], decreased drug influx as well as drug compartmentalization into lysosomes [10], and enhanced DNA repair or detoxification systems [11]. In the last decades, even the deregulation of energy metabolism has been described as a hallmark of cancer and drug resistance. This phenomenon is also called metabolic reprogramming and it involves several alterations in metabolic pathways, such as increased glycolysis, a shift toward the pentose phosphate pathway, as well as enhanced lipid synthesis, higher exploitation of glutamine pathway, and mitochondrial alterations [12]. Several studies highlighted the central role that mitochondria play in cancer development and progression [13], and in the onset of drug resistance, suggesting that targeting the mitochondrial pathway could represent a promising strategy to overcome drug resistance [7].
Previous studies in our laboratory already identified that mitochondrial alterations are strictly linked with chemotherapy resistance. In particular, we demonstrated that ovarian and osteosarcoma cisplatin-resistant cancer cells presented defects in mitochondrial morphology, in particular reduced mitochondrial mass, suggesting an enhanced activation of mitophagy and a central role played by mitochondria in cisplatin resistance [14,15]. Similarly, in DXR-resistant cells with a multidrug-resistant (MDR) phenotype, we also demonstrated a lower mitochondrial activity [16].
In light of these considerations, our hypothesis is that DXR resistance in osteosarcoma cell models could be driven also by mitochondrial dysfunctions. The purpose of this paper is to characterize the mitochondrial phenotype of two osteosarcoma cell lines, MG63 and HOS, analyzing both sensitive and DXR-resistant clones with the aim of identifying alterations in resistant clones that could be exploited for cancer therapy.

Cell Lines
The HOS and MG63 cell lines were kindly provided by IRCCS Istituto Ortopedico Rizzoli (IOR, Bologna, Italy), and derived from human osteosarcoma. The cell lines included in our research are commercially available; for this reason, it is not necessary to obtain ethical approval. The DXR-resistant variants HOS DXR 10 ng/mL, HOS DXR 30 ng/mL, HOS DXR 100 ng/mL, as well as MG63 DXR 30 ng/mL and MG63 DXR 100 ng/mL, were obtained by continuous exposure to increasing DXR (Teva Pharmaceutical Industries, Tel Aviv, Israel) concentrations, as reported by Avnet and coworkers [10]. Approximately 2-3 weeks (4-5 passages) were required to establish adequate growth at each DXR concentration. The cells were grown in Iscove's Modified Dulbecco's Medium (IMDM) medium (Corning, New York, NY, USA) supplemented with 10% fetal bovine serum (FBS) (Thermo Fisher Scientific, Waltham, MA, USA), 4 mM glutamine (Corning, New York, NY, USA), 100 U/mL penicillin, and 100 µg/mL streptomycin (Corning, New York, NY, USA), in humidified condition at 5% CO 2 and 37 • C. Cells were collected every 2 days with the minimum amount of 0.25% trypsin-0.2% EDTA (Euroclone, Milan, Italy).

Western Blotting
The Lowry method was used to evaluate the protein content after the cells had been seeded in 6-well plates, cultured in complete media, and lysed using ice-cold lysis buffer supplemented with protease inhibitor cocktails (Roche Molecular Biochemicals, Mannheim, Germany) (Biorad DC Protein Assay, MA, USA). In the running buffer, 25 µg of protein from each sample was placed onto a polyacrylamide gel and electrophoretically separated. The proteins were blotted onto a Hybond-P PVDF membrane following electrophoresis (Amersham Biosciences, Buckinghamshire, UK). The membrane was exposed to anti-TOM20 (mouse, 1:1000; AbCam, Cambridge, UK), anti-VDAC1 (mouse, 1:1000; AbCam, Cambridge, UK), and anti-BNIP3 (rabbit, 1:1000; AbCam, Cambridge, UK) after blocking with a 10% skim milk solution. Following washing, the membrane was incubated with HRPconjugated anti-rabbit secondary antibody (1:3500; PerkinElmer, MA, SUA) or anti-mouse secondary antibody (1:10000; PerkinElmer, MA, USA). According to the manufacturer's instructions, the signal was seen using an improved chemiluminescent kit from Amersham Biosciences, and then it was examined using a Molecular Imager VersaDoc MP 4000 (Biorad, Hercules, CA, USA). Proteins were normalized to β-ACTIN (mouse, 1:7000, AbCam, Cambridge, UK) and GAPDH (rabbit, 1:2000, Cell Signaling, Danvers, MA, USA).

Mitochondrial ROS Levels by Flow Cytometry
A constant number of cells were seeded, and after 48 h, 5 µM Mitosox (Invitrogen, Waltham, MA, USA) was added and incubated for 30 min at 37 • C. Cells were then washed in PBS solution, detached using 0.05% trypsin and 0.02% EDTA, centrifuged for 10 min at 2000 rpm, and resuspended in a solution of PBS, 2% FBS, and 0.02% sodium azide (Sigma-Aldrich, St. Louis, MO, USA). Cells were then examined directly using a 488 Argon laser-equipped Epics XL Coulter Systems (Beckman Coulter, Brea, CA, USA). Dead cells were electronically gated out using forward versus side scatter profiles, and a minimum of 10 4 cells of interest were then examined further. Analysis using the EXPO 32 software produced the percentages of the fluorescent cells' values (Beckman Coulter, Brea, CA, USA).

Apoptosis Detection by Flow Cytometry
Cells were seeded in 6-well plates and grown in complete medium, washed with PBS, detached with 0.25% trypsin-0.2% EDTA and centrifuged for 5 min at 1200 rpm. Cells were stained with FITC Annexin V (InVitrogen, Waltham, MA, USA) and Propidium Iodide according to manufacturer's instructions, and then diluted with Annexin V-binding buffer (InVitrogen, Waltham, MA, USA). Fluorescence intensity was analyzed using BD FACS Aria instrument.

Trypan Blue Assay
Cells were plated on 12-well plates and, following overnight incubation, were exposed to different rotenone treatments (0.01-0.1-1 µM) according to the experimental protocol. After treatments, cells were washed, detached with 0.25% trypsin-0.2% EDTA, and suspended in trypan blue (Sigma-Aldrich, St. Louis, MO, USA) at a 1:1 ratio in a medium solution. Cells were counted using a chamber Burker hemocytometer.

Hoechst Cell Count in Hypoxic Conditions
Cells were plated on 96-well plates and, following overnight incubation, were exposed to 100 µM cobalt chloride (Sigma-Aldrich, St. Louis, MO, USA). 24 and 48 h from incubation with cobalt chloride, cells were stained with Hoechst 33258 2.25 µg/mL (Thermo Fisher Scientific, Waltham, MA, USA) for 20 min. To induce hypoxia from oxygen deprivation, cells were plated on 96-well plates and, following overnight incubation, were moved to the Environmental Chamber of ImageXpress Pico Automated Cell Imaging System (Molecular Devices, San Jose, CA, USA) at 2% O 2 concentration. For both the chemically induced hypoxia and hypoxia from oxygen deprivation, cells were stained with Hoechst 33258 2.25 µg/mL for 20 min at different time points. Images were acquired by fluorescent microscopy ImageXpress Pico Automated Cell Imaging System with a 10× objective. Analysis software detected the total number of cells based on Hoechst positive nuclei.

Statistical Analysis
The statistical analysis of experimental results was performed using GraphPad Prism version 7 (GraphPad, San Diego, CA, USA.). An appropriate post-hoc test (Bonferroni, Tukey-Kramer, or Šídák multiple comparison tests) was used after the one-way or twoway analysis of variance (ANOVA) of the data. For the purpose of assessing statistical significance, a p-value of less than 0.05 was taken into account.

DXR-Resistant Clones Demonstrate Changes in Mitochondrial Morphology and Loss in Mitochondrial Membrane Potential
Mitochondrial morphology can be modified in response to environmental and cellular stresses, such as DXR treatment. This process is regulated by fusion and fission dynamics and plays a pivotal role in cellular activity [18].
The evaluation of mitochondrial network of sensitive and resistant osteosarcoma cell lines by confocal microscopy was performed through the orange-fluorescent probe Mitotracker Orange, which stains mitochondria with a membrane potential-dependent accumulation. Figure 1a shows the mitochondrial network of sensitive MG63 and resistant clones, DXR 30 ng/mL, and DXR 100 ng/mL, while Figure 1b refers to HOS and relative DXR-resistant clones, DXR 10 ng/mL, DXR 30 ng/mL, and DXR 100 ng/mL. Images suggested no difference in the mitochondrial network, which is similar and filamentous in both sensitive and resistant cancer cells. However, the membrane potential is lower in resistant clones compared to sensitive cells.
To confirm the data obtained by confocal microscopy, the mitochondrial membrane potential of sensitive and resistant osteosarcoma cells was evaluated in both MG63 and HOS cell lines through flow cytometry analysis. Cells were stained with Rhodamine 123, a fluorescent probe sequestered by active mitochondria in a membrane potential fashion. Figure 1c,d shows that all resistant clones significantly reduced membrane potential compared to sensitive clones, thus confirming results obtained with confocal microscopy images.
These preliminary results suggest an alteration in the mitochondrial status of DXRresistant osteosarcoma cells, indicating that further investigations are needed to understand the molecular mechanisms responsible for these changes.

The Mitochondrial Mass Was Lower in DXR-Resistant Clones than in Sensitive Cells: Is Mitochondrial Biogenesis or Mitophagy the Involved Pathway?
Mitochondria are dynamic organelles and changes in mitochondrial mass reflect unbalance between mitochondrial biogenesis and mitophagy [19]. Increased or decreased mitochondrial mass has been correlated with cancer [20]. Nonyl acridine orange (NAO), a fluorescent probe binding cardiolipin, was used to assess the mitochondrial mass of sensitive and DXR osteosarcoma cells by flow cytometry. Results showed a significant decrease of NAO fluorescence percentage in all resistant clones compared to sensitive cancer cells (Figure 2a    To confirm these data, we even assessed the protein expression of VDAC1 and TOM20, which are already considered specific markers of mitochondrial mass. As it is possible to observe in Figure 2c,d, the expression of both proteins is significantly decreased in resistant cells with respect to their sensitive counterparts. The uncropped Western Blot images can be found in Figures S1 and S2. Considering that the equilibrium between mitochondrial biogenesis and mitophagy regulates mitochondrial mass [21], we analyzed some genes involved in both these pathways to dissect which is the process involved in mitochondrial alteration. For mitochondrial biogenesis, we checked the mRNA levels of PGC1-α, TFAM, and NRF-2. As Figure 3a,b shows, the expression of both PGC1-α and TFAM decreases in MG63 DXR-resistant cells, and TFAM levels are lower in HOS DXR-resistant clones. These results suggest that reduced biogenesis could be responsible for the observed reduction of mitochondrial mass in resistant clones. As further confirmation, the protein expression on BNIP3, a marker of mitophagy [22], downregulated in the MG63-DRX resistant clones, thus excluding the involvement of mitophagy as the process responsible for reduced mitochondrial mass ( Figure 3c). The uncropped Western Blot images can be found in Figure S2.

DXR Osteosarcoma Cells Show Decreased mtROS Levels and Different Response to Hypoxia
Cellular homeostasis is also regulated by the oxidative status, and alterations such as an increase in ROS production represent a factor involved in cell death. In the context of chemotherapy, it is known that the mode of action of several chemotherapeutic agents, such as cisplatin and DXR, is related to oxidative stress induction, which leads to cell death. On the other side, one of the mechanisms of drug resistance is the increased antioxidant system, which counteracts the enhanced chemotherapy-induced oxidative stress [23], and that might be obtained by reducing the biogenesis ROS-producing mitochondria. Therefore, we analyzed the ROS content. As confirmation, results in Figure 4 a,b showed Furthermore, we evaluated the mRNA levels of BNIP3 in HOS cell lines, as reported in Figure 3d. The results showed that only HOS DXR 30 ng/mL cancer cells slightly increase the expression, but not in a significant way.

DXR Osteosarcoma Cells Show Decreased mtROS Levels and Different Response to Hypoxia
Cellular homeostasis is also regulated by the oxidative status, and alterations such as an increase in ROS production represent a factor involved in cell death. In the context of chemotherapy, it is known that the mode of action of several chemotherapeutic agents, such as cisplatin and DXR, is related to oxidative stress induction, which leads to cell death. On the other side, one of the mechanisms of drug resistance is the increased antioxidant system, which counteracts the enhanced chemotherapy-induced oxidative stress [23], and that might be obtained by reducing the biogenesis ROS-producing mitochondria. Therefore, we analyzed the ROS content. As confirmation, results in Figure 4a,b showed ROS decrease in DXR-resistant osteosarcoma cells compared to sensitive counterparts.   The reduced mitochondrial mass and biogenesis and the reduced ROS production in resistant cells suggest that these cells have reduced mitochondrial oxidative phosphorylation. To validate this hypothesis, we measured the cellular response to low-oxygen (2%) or chemically induced hypoxia (with CoCl 2 ). CoCl 2 stabilizes hypoxia-inducible factors 1α and 2α under normoxic conditions and simulates a hypoxic microenvironment [24]. As confirmation of the hypoxia-like metabolic state in CoCl 2 -treated cells, mRNA expression levels of the indirect hypoxia marker CAIX [25] were induced in all cells tested ( Figure A1). We then counted the number of cells after incubation with both low oxygen tension and CoCl 2 . Before 144 h, cell growth was mostly unaffected by both types of hypoxia, with the exception of sensitive cells treated with CoCl 2 ( Figure A2). In contrast, after longer exposure (at 144 h under low O 2 tension), all cell types were significantly affected compared with the normoxic condition (Figure 5c-j). Overall, it can be concluded that resistant cells appear to be much less dependent on oxygen metabolism than sensitive cells, as attested by both CAIX expression levels and cell numbers at 48 h when treated with CoCl 2 (Figure 5a,b).

DXR-Resistant Osteosarcoma Cells Show Different Behavior under Metabolic Stresses
DXR-resistant osteosarcoma cells revealed a different mitochondrial profile with respect to their sensitive counterpart. Accumulating evidence showed that mitochondrial dysfunctions can be highlighted when oxidative metabolism is increased, such as glucose deprivation and galactose addition in the medium. The next step of this work was to evaluate the response of sensitive and DXR-resistant cancer cells under two metabolic stresses, glucose, and glutamine deprivation. MG63 DXR-resistant clones significantly reduced cell viability after 24 h in both glucose-free/galactose 5 mM, and glutamine-free media (Figure 6a,b). On the other side, among the resistant HOS clones, only HOS DXR 100 ng/mL cells resulted to be affected by these metabolic stresses (Figure 6c,d).
These results, taken together, suggest mitochondrial dysfunctions in DXR-resistant cancer cells, which are more dependent on glucose and glutamine pathways with respect to sensitive counterparts.

Targeting Mitochondrial Biogenesis to Restore DXR Sensitivity
Our results highlighted a different mitochondrial profile of DXR-resistant osteosarcoma cells compared to sensitive cells, in particular demonstrating reduced mitochondrial membrane mass, potential, and activity. Further investigations assessed that the process responsible for decreased mitochondrial mass in resistant cells is likely the reduction of mitochondrial biogenesis. In line with these observations, we reasoned to verify whether the induction of mitochondrial biogenesis was able to re-sensitize DXR effect in osteosarcoma-resistant clones. The previous literature reported that the natural compound quercetin is an inducer of mitochondrial biogenesis [26]. We evaluated cell viability of both sensitive and DXR-resistant cancer cells following 24, 48, and 72 h of the combined treatment with quercetin 0.1-200 µM and DXR, at the respective concentration of each resistant cancer cell lines (DXR 10-30-100 ng/mL). Results showed that the association with quercetin re-sensitized the effect of DXR in MG63 and HOS DXR-resistant cancer cells (Figures 7 and 8, Table 1, Table 2, Tables S1 and S2), by enhancing the percentage of late apoptotic cells ( Figure A3). In addition, we also demonstrated that quercetin 10 µM enhanced TFAM expression following 72 h of treatment ( Figure A4). DXR-resistant osteosarcoma cells revealed a different mitochondrial profile with respect to their sensitive counterpart. Accumulating evidence showed that mitochondrial dysfunctions can be highlighted when oxidative metabolism is increased, such as glucose deprivation and galactose addition in the medium. The next step of this work was to evaluate the response of sensitive and DXR-resistant cancer cells under two metabolic stresses, glucose, and glutamine deprivation. MG63 DXR-resistant clones significantly reduced cell viability after 24 h in both glucose-free/galactose 5 mM, and glutamine-free media ( Figure  6 a,b). On the other side, among the resistant HOS clones, only HOS DXR 100 ng/mL cells resulted to be affected by these metabolic stresses (Figure 6 c,d). These results, taken together, suggest mitochondrial dysfunctions in DXR-resistant cancer cells, which are more dependent on glucose and glutamine pathways with respect to sensitive counterparts.

Targeting Mitochondrial Biogenesis to Restore DXR Sensitivity
Our results highlighted a different mitochondrial profile of DXR-resistant osteosarcoma cells compared to sensitive cells, in particular demonstrating reduced mitochondrial membrane mass, potential, and activity. Further investigations assessed that the process responsible for decreased mitochondrial mass in resistant cells is likely the reduction of mitochondrial biogenesis. In line with these observations, we reasoned to verify whether the induction of mitochondrial biogenesis was able to re-sensitize DXR effect in osteosarcoma-resistant clones. The previous literature reported that the natural compound quercetin is an inducer of mitochondrial biogenesis [26]. We evaluated cell viability of both sensitive and DXR-resistant cancer cells following 24, 48, and 72 h of the combined treatment with quercetin 0.1-200 M and DXR, at the respective concentration of each resistant cancer cell lines (DXR 10-30-100 ng/mL). Results showed that the association with quercetin re-sensitized the effect of DXR in MG63 and HOS DXR-resistant cancer cells ( Figure  7,8, Table 1,2, Table S1,S2), by enhancing the percentage of late apoptotic cells ( Figure A3).

Discussion
Mitochondria play a pivotal role in cell life, coordinating several signaling processes such as the tricarboxylic acid (TCA) cycle, fatty acid oxidation (FAO), and oxidative phosphorylation (OXPHOS) [27], and regulating energy homeostasis and cell death via apoptosis. Recent studies correlated mitochondria and cancer, suggesting that mitochondrial dysfunctions are responsible for tumor development and metastasis progression and also for alterations in chemotherapy's response. Evidence reported that several chemotherapeutic agents accumulate into mitochondria, thus causing mtDNA damage [28]. These organelles are considered one of the main targets of DXR, through abnormalities in mitochondrial shape and functionality and mitochondrial death mediated by apoptosis.
Considering the pivotal role of mitochondria in tumor development and in DXR pharmacological activity, this work aimed to characterize the mitochondrial phenotype of sensitive and DXR-resistant osteosarcoma cells. Once alterations were identified, the goal was to target them to restore drug sensitivity in resistant cells.
Several studies reported that alterations in the mitochondrial network, membrane potential, and mass are correlated with chemoresistance or cancer aggressiveness in different types of tumors such as colon, breast, and gastric cancers [29][30][31]. These aspects reflect the functional and healthy state of these organelles and could become indicators of an unhealthy and dysfunctional status. Our results showed that DXR-resistant cancer cells decreased both the mitochondrial number and the membrane potential, although their morphology remains unchanged. The mitochondrial membrane potential (∆Ψ m ) is a thorough indicator of mitochondrial function, and its dysregulation causes an increase in reactive oxygen species production, with consequent oxidative damage to DNA and other cellular structures, and ultimately results in cell death [32,33]. It is interesting to note that several cell models have shown a decrease in mitochondrial membrane potential after exposure to DXR in vitro [34], and our data are in line with these findings. In fact, all DXR-resistant osteosarcoma cells showed a reduced mitochondrial membrane potential with respect to the sensitive counterpart.
It is known that DXR is a redox-active substance that can interact with a number of enzymes, such as mitochondrial complex I, to produce large quantities of ROS and the semiquinone radical [35]. Our results indicate that mitochondria of resistant cells present lower levels of mitochondrial ROS production. Furthermore, DXR-resistant osteosarcoma cells also showed a reduction of mitochondrial mass, confirmed both with NAO fluorescence and the decreased expression of two mitochondrial markers, VDAC1 and TOM20. It is known that mitochondrial mass is regulated by the equilibrium between mitochondrial biogenesis and mitophagy. Alterations of these processes have been correlated with chemoresistance. Peroxisome proliferator-activated receptor-coactivator 1 (PGC-1) and mitochondrial transcription factor A (TFAM), two mitochondrial transcription factors, mediate an increase in mitochondrial number in response to cellular injury [36]. According to several studies, the mitochondrial biogenesis pathway's genes and proteins are upregulated, and these changes cause cancer chemoresistance. Increased expression of TFAM and PGC-1α was assessed in tumors unresponsive to cisplatin and sorafenib; it was also demonstrated that the depletion of these proteins restored the sensitivity to the drugs [37][38][39]. On the other side, a recent study in our laboratory demonstrated that the increased expression of BNIP3, which is a protein involved in the mitophagy process, is related to cisplatin resistance. Both genetic and pharmacological approaches through inhibition showed a re-sensitization of resistant cancer cells to the chemotherapeutic agent, supporting the idea that targeting mitochondrial dynamics could be a valuable strategy in chemosensitization. Interestingly, our data demonstrated that DXR-resistant osteosarcoma cells decreased mRNA levels of TFAM gene compared to their sensitive counterpart, thus suggesting that the downregulation of mitochondrial biogenesis could be the process involved in the DXR resistance onset. In line with this, the evaluation of the mitophagic marker BNIP3 shows that resistant cells didn't present significant alterations, prompting us to exclude an increased mitophagy as the mechanism responsible for a reduced mitochondrial mass. This result is not very surprising because in previous studies, we have already observed that osteosarcoma cells do not tend to activate autophagy as an adaptation mechanism to stressful conditions [10,40].
Having identified the downregulation of mitochondrial biogenesis genes as a possible mechanism of resistance, we sought to combine DXR treatment with a drug that may induce the activation of this process and thus re-sensitize DXR-resistant cells. Quercetin is a natural flavonoid already known to be an inducer of mitochondrial biogenesis [41]. In our DXR-resistant cell model, combining DXR with quercetin re-sensitized the resistant cells to DXR. However, quercetin is also a monocarboxylate transporter (MCT) inhibitor [42]. Since MCTs play a significant role in maintaining intracellular pH by transporting excess protons outside cancer cells, it is also possible that quercetin treatment acidifies the cytosol in resistant cells. The subsequent increase of the pH gradient at the plasma membrane may facilitate the cellular intake of DXR, which is a weakly basic molecule, and thus increase its intracellular accumulation and toxicity [10]. However, in this study, we also showed that quercetin treatment increases TFAM levels in MG63-resistant cells, confirming that, in these cells, quercetin treatment has the potential to restore DXR sensitivity by inducing mitochondrial biogenesis. In conclusion, our preliminary results pave the way for TFAM inducers or other mitochondrial inducers as possible DXR sensitizers. The combination of compounds acting on different targets represents a promising strategy in the context of chemoresistance therapy because it allows for a reduction in the dosage of single drugs, thus alleviating severe toxicities and at the same time enhancing the therapeutic effectiveness.

Conclusions
Taken together, these data highlighted a different mitochondrial profile between sensitive and DXR-resistant osteosarcoma cells. In particular, resistant clones showed a decreased mitochondrial mass and reduced expression of mitochondrial biogenesis genes. The phenotyping of the mitochondrial profile allows the identification of altered pathways, which can be exploited by novel pharmacological approaches that selectively target mitochondrial dysfunctions in resistant cancer cells. In this work, the treatment of osteosarcoma cells with quercetin, a natural compound able to induce mitochondrial biogenesis, caused a re-sensitization of DXR-resistant cells to the chemotherapeutic agent. These preliminary results open the window to the development of promising pharmacological strategies to overcome DXR resistance in osteosarcoma.  between treated and untreated. (d) Representative cytograms of apoptotic cells analyzed with Annexin V-PI staining. HOS, HOS DXR 10 ng/mL, HOS DXR 30 ng/mL, and HOS DXR 100 ng/mL cells were treated 72 h with quercetin 10 M. The graphs represent the quantification of early (e) and late apoptosis (f) by showing the ratio between treated and untreated. Data are shown as mean ± SD of three different experiments. *p < 0.05, treated "vs." untreated cells.