MT2A is an early predictive biomarker of response to chemotherapy and a potential therapeutic target in osteosarcoma

Osteosarcoma is the most prevalent primary bone malignancy in children and young adults. Resistance to chemotherapy remains a key challenge for effective treatment of patients with osteosarcoma. The aim of the present study was to investigate the preventive role of metallothionein-2A (MT2A) in response to cytotoxic effects of chemotherapy. A panel of human and murine osteosarcoma cell lines, modified for MT2A were evaluated for cell viability, and motility (wound healing assay). Cell-derived xenograft models were established in mice. FFPE tumour samples were assessed by IHC. In vitro experiments indicated a positive correlation between half-maximal inhibitory concentration (IC50) for drugs in clinical practice, and MT2A mRNA level. This reinforced our previously reported correlation between MT2A mRNA level in tumour samples at diagnosis and overall survival in patients with osteosarcoma. In addition, MT2A/MT2 silencing using shRNA strategy led to a marked reduction of IC50 values and to enhanced cytotoxic effect of chemotherapy on primary tumour. Our results show that MT2A level could be used as a predictive biomarker of resistance to chemotherapy, and provide with preclinical rational for MT2A targeting as a therapeutic strategy for enhancing anti-tumour treatment of innate chemo-resistant osteosarcoma cells.


Results
MT2A mRNA level correlates with chemotherapy IC50 values. We first determined the expression level of MT2A in a panel of eight osteosarcoma cell lines by real-time quantitative RT-PCR. Cell lines exhibited various basal expression level of MT2A mRNA (Fig. 1A). We then determined the dose-response to five currently used chemotherapeutic drugs, namely cisplatinum, doxorubicin, etoposide, mafosfamide and methotrexate. Cells in exponential phase of growth were exposed to the indicated drugs for 72 hrs, and cell viability was assessed by the MTS test. All tested drugs dose-dependently reduced cell viability, but cell lines exhibited various sensitivity to a given drug, as reflected by half maximal inhibitory concentration (IC50) values (Supplemental Table 2). IC50 values are consistent with those referenced in 8 .
We then determined the correlation between MT2A mRNA level and IC50 values for each chemotherapeutic drug ( Fig. 1B-F). MG63 cell line was removed from the correlation analysis because of its huge basal MT2A expression. A strong positive correlation was detected between MT2A level and resistance to all tested drugs (R > 0.7, p < 0.05).
Taken together, these results indicate that MT2A mRNA level correlate with cell resistance to chemotherapy-based therapies currently used in osteosarcoma.
Chemotherapy treatment induces MT2A mRNA expression. To evaluate the up-regulation of MT2A level by chemotherapy, cells were exposed to cisplatinum, doxorubicin, etoposide, mafosfamide, methotrexate or www.nature.com/scientificreports www.nature.com/scientificreports/ solvent for 24 hrs, and MT2A mRNA level was determined by RT-qPCR. All tested drugs induced an about 2-fold up-regulation of MT2A mRNA level in all tested cell lines (Fig. 2). Then, to evaluate the possible involvement of glutathione (GSH), which is another key contributor to xenobiotic detoxication, cells were exposed to chemotherapeutic drugs for 24 hrs, and GSH level was determined using a colorimetric assay. Only etoposide induced a decrease in intracellular GSH level in all cell lines (−26 to −70%, p < 0.01 vs. solvent; Supplemental Fig. 1), whereas all other tested drugs did not affect GSH content.
Taken together, these results indicate that osteosarcoma cells rather promote MT2A neo-synthesis than trigger the glutathione-dependent detoxication in response to cisplatinum, doxorubicin, mafosfamide, or methotrexate. Etoposide exposure led to both MT2A induction and GSH efflux.
MT2A level does not correlate with osteosarcoma cell migration ability. We determined the migration speed of our panel of osteosarcoma cell lines both in basal culture conditions (Supplemental Fig. 2A), and combined them to MT2A mRNA level (Supplemental Fig. 2B). No correlation was detectable (R < 0.2; p > 0.5).
These results indicate that MT2A level did not influence cell migration rate.    Table 2). As expected, the cells modified with sh-MT2A sequences exhibited significantly lower MT2A mRNA level (−50 to −70% vs. shControl; Supplemental Fig. 4). We then evaluated the impact of lowered MT2A level on the dose-response to cisplatinum, doxorubicin, etoposide or ifosfamide. Cells in exponential phase of growth were exposed for 72 hrs to the indicated drugs. The MT2A-silenced cells exhibited a higher sensitivity to a given drug, as reflected by lower half maximal inhibitory concentration (IC50) values as compared to shControl cells (Table 1).
To evaluate if repression of MT2A can reverse chemo-resistance, we modified HOS cells that have acquired in vitro resistance to doxorubicin (Resistance Index = 88; see Methods section) 9 . This HOS-Doxo R cell line exhibited a cross-resistance to etoposide (RI = 140) but not to cisplatin or ifosfamide 9 . As expected, Hos-Doxo R cells modified with sh-MT2A sequences exhibited significantly lower MT2A mRNA level (−40% vs. shControl; Supplemental Fig. 4C). The MT2A-silenced cells exhibited a higher sensitivity to a given drug, as reflected by lower IC50 values as compared to shControl cells (Table 2).
Taken together, these results indicate that silencing MT2A expression reinforces the cell sensitivity to chemotherapeutic drugs, and can reverse cell chemo-resistance.

MT2A silencing promotes anti-tumour effects of chemotherapy in Xenograft models.
We performed similar experiments with the murine osteosarcoma cell line K7M2. Stable transduction of shRNA sequences targeting MT2 gene resulted in significant decrease in MT2 mRNA level, as compared to shControl or parental cells (−55%, p < 0.05; Fig. 3A). No significant modification in cell morphology or cell proliferation rate was detectable among the modified and parental cell lines (Fig. 3B,C, respectively). As observed in human cell lines, (i) only etoposide reduced GSH content in both shControl and shMT2 modified K7M2 cells (Fig. 3D), (ii) all tested chemotherapy drugs induced MT2 mRNA expression in parental K7M2 cells (Fig. 3E), and (iii) MT2-silenced cells exhibited a higher sensitivity to a given drug, as reflected by lower IC50 values as compared to shControl cells (Fig. 3F).
Taken together, these results indicate that the murine K7M2 cell line behaves the same as human cell lines and represents a suitable model for further in vivo investigations.
Pulmonary metastatic foci were assessed by HES staining of FFPE lung sections (Fig. 6A). As expected, the average metastatic surface was markedly reduced by chemotherapy regimens in the mice injected with shControl cells: from −60% with ifosfamide (p = 0.037 vs. saline; Fig. 6B); −61% with doxorubicin (p = 0.036); −76% with etoposide (p = 0.0042) to −87% with cisplatinum (p = 0.0031). No difference could be detected in the average metastatic surface at sacrifice between shMT2 and shControl groups without treatment (saline; p > 0.3). The average metastatic surface was not more drastically reduced in the mice injected with shMT2 cells, as compared to mice injected with shControl cells when treated with chemotherapy (p > 0.2). www.nature.com/scientificreports www.nature.com/scientificreports/ The average number of metastases was markedly reduced by chemotherapy regimens in the mice injected with shControl cells (−38% to −79%, p < 0.01; Fig. 6C), and more drastically reduced in the mice injected with shMT2 cells by cisplatinum (−96%, p = 0.00045 vs. saline) or doxorubicin (−77%, p = 0.0013), whereas no difference could be detected between shMT2 and shControl groups without treatment (saline; p = 0.335).
Overall, the metastatic surface and the number of metastases positively correlated with tumour volume (Fig. 6D,E, respectively).
Taken together, these results confirm that MT2A level influence chemotherapy induced-cell death, and primary tumour development resulting in modulation of pulmonary metastatic dissemination. www.nature.com/scientificreports www.nature.com/scientificreports/

Discussion
Chemo-resistance is a major obstacle to therapeutic success of osteosarcoma. Although neo-adjuvant chemotherapy has considerably improved the prognosis of osteosarcoma patients since the past four decades, about two out of five patients will experience relapse or/and recurrence within five years after diagnosis.
In the present study, we showed a strong positive correlation between MT2A/MT2 mRNA level and cell sensitivity to a panel of chemotherapeutic drugs currently used in clinical practice 10 , as assessed by half maximal www.nature.com/scientificreports www.nature.com/scientificreports/ inhibitory concentration (IC50) values. These data are consistent with the correlation we previously reported between MT2A mRNA level in chemo-naive biopsy, and tumour response to chemotherapy thereafter (based on Huvos scoring of tumour necrosis), and overall survival of patients 7 . Taken together, our data showed that MT2A contributes to chemotherapy resistance in osteosarcoma, and could be considered as a predictive indicator at diagnosis for osteosarcoma responsiveness to current chemotherapy regimen, whatever their mechanism of action.
We also investigated the possible role of MT2A in metastatic dissemination. We did not detect any correlation between MT2/MT2A level and in vitro cell motility using a panel of cell lines. These data are consistent with our previous study reporting no modification in cell motility or invasion capacity of MT2A-overexpressing cells or MT2A-silenced cells compared to parental cells 7 . Moreover, we did not detect any influence of MT2 silencing on primary tumour spreading to the lungs in our Cell line-Derived Xenograft (CDX) model, either in the average metastatic foci surface or number, suggesting that MT2A level may only influence cell responsiveness to toxics and not the invasion process.
In response to chemotherapeutic drugs, cancer cells could adopt several strategies to develop their resistance such as the control of the transmembrane entry/efflux of the compounds; the activation of DNA repair machinery (ERCC1 and 2, NER genes…); the adaptation of the drug target (dihydrofolate reductase (DHFR), topoi-somerase…); or the synthesis of detoxication proteins (gluthation-S-transferase (GST), metallothioneins…) for example (reviewed in 11 ). Such adaptive behaviour has been reported in osteosarcoma. Indeed, an up-regulation of DHFR and MDR1 (multidrug resistance protein 1) have been reported to be associated with an enhanced chemo-resistance to methotrexate and doxorubicin 12 . Increased expression of P-glycoprotein in tumour cells was significantly associated with poor survival in patients with osteosarcoma [13][14][15][16] . An overexpression of GSTπ, Hsp27, LRP (lung resistance-related protein) as well as MTs has been detected in osteosarcoma tumours with poor histologic response to pre-operative chemotherapy 17 . However, no statistical differences could be detected in terms of response to treatment and patient outcome. This apparent contradiction with our data could be related to the specificity of the molecular tools. Almost all studies focused on an immuno-histochemical detection of MTs in FFPE tissue samples. However, the MTs antibodies that are commonly used for such IHC staining do not distinguish between the different highly homologous isoforms of the MT superfamily. This could minimize the impact of the variations of one or a few of the isoforms.
In our panel of osteosarcoma cell models, IC50 values were heterogeneous from one cell line to another, but overall correlated to the MT2/MT2A mRNA level. We also observed that all tested drugs induce MT2/MT2A expression, independently of the genomic alterations detectable in the cell lines. Such adaptive process could explain why tumours become resistant as cycles of chemotherapy progress. The evaluation of MT2A mRNA level would inform about the innate resistance status when measured at diagnosis, and about the acquired resistance linked to an adaptive process if monitored during treatment.
We also focused here on the possible complementary role of the glutathione pathway in the observed MT2A-related chemo-resistance. GST catalyses the conjugation of toxic electrophilic chemicals to the tri-peptide glutathione (GSH) in order to facilitate their rapid clearance from the cell 18 . Increased GST expression has been associated with a resistant phenotype to conventional chemotherapy in different cancer types (reviewed in 19 ). Some GST polymorphisms are correlated to clinical outcome of osteosarcoma patients 20 . In our panel of osteosarcoma cell lines we did not observe any variation in intracellular GSH levels after exposure to chemotherapy with the exception of a decrease for etoposide, that is known to undergo glutathione conjugation 21 and to increase GSH export 22 . In addition, the GSH levels are comparable in cells modified with non-relevant shRNA sequences or silenced for MT2/MT2A, even under chemotherapy treatment. Taken together, these data are consistent with our hypothesis that the detoxication process triggered by osteosarcoma cells incubated in the presence of chemotherapeutic drugs is mainly metallothionein-dependent.
The second significant finding of our study is the proof of concept that MT2A could also be a valuable therapeutic target to improve osteosarcoma sensitivity to chemotherapy. We showed that MT2/MT2A silencing enhanced cytotoxic action of current chemotherapeutic drugs on osteosarcoma cells in vitro, reflected by the lowering of IC50 values for all tested drugs. More importantly, we showed that MT2 silencing led to higher cytotoxic effects of drugs in preclinical Cell line-Derived Xenograft (CDX) models. This fully confirmed our hypothesis that MT2 targeting rendered tumours significantly more sensitive to chemotherapy. Taken together, these data show that MT2A could be considered as a therapeutic target to prevent or reduce osteosarcoma resistance to chemotherapy. Not many strategies of targeting MT isoforms have reached the clinical practice as therapeutic agents (reviewed in 6 ). Further investigations are required to identify a physiological and clinically safe strategy to prevent MT2A neo-synthesis under treatment or to reduce MT2A levels in case of high basal content.
In summary, this study identified the key role of high MT2A expression level in the resistance of osteosarcoma cells to chemotherapeutic drugs. In effect, we revealed a negative correlation between MT2A level and drug efficacy (IC50 values), and anti-tumour effects for various chemotherapeutic drugs when cells were silenced for MT2A. Our results thus support a model whereby MT2A contributed to lowering the drug cytotoxic effects and could be a valuable therapeutic target to improve osteosarcoma sensitivity to chemotherapy. In conclusion, we demonstrated that MT2A could be considered as a predictive biomarker of the efficacy of chemotherapy for patients with osteosarcoma, and as a potential therapeutic target to develop novel treatment strategy to prevent or de-escalate chemo-resistance in osteosarcoma. All the cells were cultured in Dulbecco's Modified Eagle's Medium (DMEM) supplemented with 10% heat inactivated Foetal Calf Serum, at 37 °C in a 5% CO2 humidified atmosphere. The culture media were replaced three times a week. Lentiviral transduction. Cells were seeded at a density of 25,000 cells/cm². The next day, cells were transduced with lentiviral particles encoding human sh-MT2A, murine sh-MT2 or non-targeting (control) shRNA sequences purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA), according to the manufacturer's instructions. Virus-containing medium was changed with fresh medium 24 hrs after infection. Four days later, stably modified cells were selected with 10 µg/ml puromycin dihydrochloride (Sigma-Aldrich) for 3 days. Viable cells were then routinely maintained in complete medium.

Methods
In vitro cell viability assay. Cells were seeded at a density of 15,000 cells/cm², and the next day treated with the indicated drug for 24 or 72 h. Cell viability was evaluated by the MTS tetrazolium assay (CellTiter 96 aqueous one solution cell proliferation assay kit; Promega, Charbonnieres, France), according to the manufacturer's instructions.
Wound healing assay. The cell migration rate was assessed using the IncuCyte Live Cell Imaging system according to the manufacturer's instructions (Essen Bioscience Ltd., Birmingham, UK). Cells monolayers were scratched, then washed once and maintained in culture with phase-contrast photographs collected every 4 hrs for 72 hrs.
Glutathione assay. The level of total glutathione (GSSG + GSH) was assessed using glutathione assay kit (Sigma-Aldrich), according to the manufacturer's recommendations.
RNA extraction. The RNA was extracted using TRIzol Reagent (Thermo Fisher Scientific, Illkirch, France) according to the manufacturer's protocol, and stored at −80 °C in RNase-free water. Purity and quantity of RNA samples were assessed using the NanoVue Plus spectrophotometer (GE Healthcare Life Sciences, Velizy-Villacoublay, France).
Quantitative PCR was performed on ViiA7 real-time PCR system (Thermo Fisher Scientific) using SYBR Green Master kit (Thermo Fisher Scientific) supplemented with 0.5 µM of specific primers (Supplemental Table 1). Thermal conditions were: activation for 15 min at 95 °C then 40 cycles of denaturation at 95 °C for 20 sec, 59 °C annealing for 15 sec and 72 °C extension for 15 sec. Upon completion of each run, a melting curve analysis was performed to check specificity of the primers. All signals with threshold cycle (Ct) > 36 were set as undetermined. The relative gene expression was calculated using the 2 −∆∆Ct method.
Xenograft models. All procedures were conducted according to the guidelines formulated by the European Commission for experimental animal use (L358-86/609EEC), and with the approval of the French local ethical animal committee (Comité d' éthique en Expérimentation Animale n°26/CEEA26, Villejuif, France). Severe combined immunodeficiency (SCID) female mice (Charles River, Arbresle, France) of 5-weeks old were randomized, housed at 22 °C to 24 °C with a 12-hour light/dark cycle and maintained in a pathogen-free enriched environment throughout the experiment. They were fed with a standard mice pellet diet and had free access to water. After acclimatization for 7 days, mice were injected intramuscularly (IM) with K7M2 cells (10 6 cells/15 µl PBS) in both thighs under isoflurane/air inhalational anaesthesia. After 10 days allowing tumour growth, mice were treated by intraperitoneal injection of cisplatinum (7 mg/kg body weight), doxorubicin (4 mg/kg body weight), etoposide (5 mg/kg body weight) or ifosfamide (100 mg/kg body weight) every 5 days. A vehicle-treated (saline) group was included. At day 28, mice were killed by CO 2 asphyxiation. Tumour diameters were measured with digital callipers, and the tumour volume in mm 3 was calculated by the formula = width × height × length × (π/6). Tumours and lungs were fixed in 4% paraformaldehyde in PBS before paraffin embedding.
In situ cell death detection was performed on tumour sections by TUNEL (terminal deoxynucleotidyl transferase-mediated dUTP nick-end labelling) assay according to the manufacturer's recommendation (Roche Diagnostics, Boulogne-Billancourt, France). Slides were revealed with permanent Red (Dako-Agilent Technologies, Courtaboeuf, France) and counterstained with haematoxylin.
Immuno-histological detection of MTs and Ki67 was performed on tumour sections. After immersion in 3% H 2 O 2 for 10 min to block endogenous peroxidase activity, and an antigen retrieval heating step, with a citrate buffer (pH 6), at 98 °C for 30 min, slides were incubated for 10 min with blocking reagent (Klear Mouse HRP; GBI Labs, Bothell, WA, USA) to block nonspecific staining, and incubated for 60 min with mouse monoclonal anti-mouse metallothionein antibody (clone E9, Dako-Agilent Technologies) at a dilution of 1:200, or with anti-Ki67, 1:200 (Neomarkers, LabVision, Microm Microtech, Francheville, France). After two rinses in wash buffer, slides were