OS cell lines were purchased from ATCC. Specific media requirements have been previously described for LM7 (28), OS482 (9), and MG63 and MG63.3 (29, 30). K7M2 was grown in DMEM with 10% FBS and 1% PS; HOS and HOS-MNNG were grown in EMEM with 10% FBS and 1% PS; U2OS and SaOS2 were grown in McCoy’s 5A Medium with 10% FBS and 1% PS; hFOB 1.19 (hFOB) was grown in DMEM (without phenol red) with 10% FBS and 1% PS; and 143B was grown in DMEM with 10% FBS, 1% PS, and bromodeoxyuridine (0.015 mg/mL).

PDX lines were provided by John Healey’s lab at Memorial Sloan Kettering Cancer Center (MSKCC). OS24, OS33, and OS69 were grown in RPMI with 10% FBS and 1% PS, while OS29 was grown in DMEM with 20% FBS and 1% PS. All cell lines were maintained at 37°C in an incubator supplied with 5% CO₂.

To measure glutamate release by cultured cells, 6,000 cells/plate were seeded in 3.5-cm plates using glutamine-free media for all cell lines except OS69 and OS33 (for which 12,000 cells/plate were seeded). The media was collected for seven consecutive days, and cell viability was measured using trypan blue staining. At the end of the seventh day, a glutamate assay was performed using a glutamate assay kit from Abcam (#ab83389). In a 96-well reaction plate, 2 µL of the media that had been collected each day was added to wells in duplicate. In each reaction well, 138 µL glutamate assay buffer, 8 µL glutamate developer, and 2 µL glutamate enzyme mix were added and the plate was incubated at 37° C for 30 minutes while protected from light. The absorbance was measured at 450 nm using a spectrophotometer (SpectraMax i3).

Growth inhibition was measured using an MTT (3-(4,5-dimethylthiazol-2-yl]-2,5 diphenyl tetrazolium bromide) assay. Twenty thousand cells/well were seeded in 24-well plates for all cell lines except OS69 (40,000 cells/well). After 24 hours, the media was changed and the cells were treated with either dimethylsulfoxide (DMSO), 5 µM riluzole, 10 µM riluzole, 25 µM riluzole, 50µM riluzole or 100 µM riluzole in quadruplicate. The cells were incubated for 48 h at 37°C and MTT reagent was added to cells at 5 mg/mL concentration. The cells were incubated for 2 h at 37°C, and the media was discarded; 1 mL of DMSO was then added to each well to dissolve the precipitated MTT crystals. The plate was incubated for 10 min at room temperature protected from light, and the absorbance was measured by at 570 nm with a reference wavelength at 670 nm using a spectrophotometer. The assay was independently conducted three times for each cell line.

Quantitative polymerase chain reaction (qPCR) was performed by seeding 0.6 million cells in a 100-mm plate. After 24 h, RNA was extracted following the protocol from the RNeasy Mini Kit (QIAGEN #74106). The RNA concentration was measured using Nanodrop One, and 1 µg of RNA was used for reverse transcription using a QuantiTect Reverse Transcription Kit (QIAGEN #205313). The cDNA was further diluted at 1:10 and used for qPCR. Primers were used at 4 µM concentration. SYBR Green PCR Master Mix (Thermofisher #4312705) was used as the fluorescent probe and GAPDH as the control. The following specific primers were used: Sequences of primers for GAPDH, forward: TGCACCACCAACTGCTTAGC; reverse: GGCATGGACTGTGGTCATGAG. mGluR5, forward: ATGACGGTGAGAGGTCTGCTGA; reverse: GATGCCACCAACAGCTTCTCG. mGluR1, forward: CTGGCATGAAGGAGTGCTGAAC; reverse: GCAGCTCACTTCTCCTTTCCG. Human MMP2, forward: TTGACGGTAAGGACGGACTC; reverse: ACTTGCAGTACTCCCCATCG. Human MMP9, forward: GAGACCGGTGAGCTGGATAG; reverse: TACCGAACGGTGAAGGTGAG.

For western blotting, we used antibodies for p53 (Proteintech #10442-1-AP), mGluR1 (Abclonal #A11462), mGluR5 (Santa Cruz Biotechnology #sc-293442), vGLUT1 (Abcam #ab272913), xCT (Abcam #ab307601), β-actin anti-mouse 8H10D10 (Cell Signaling Technologies #3700S), and GAPDH anti-rabbit (Cell Signaling Technology #5174S). The secondary antibodies conjugated to horseradish peroxidase were purchased from Cell Signaling Technologies. The western blot signal was detected using C-Digit from LiCOR.

Cells were seeded in a 24-well plate (40,000–50,000/well). After 24 h, the cells were treated with riluzole (50 or 100 µM) and incubated for 1 h; DMSO was used as control. After 1 h, media from the wells was discarded and PBS containing the fluorogenic probe 2’7’-dichlorodihydrofluorescin diacetate (DCFH-DA) was added to the treated and untreated wells. Next, 50 or 100 µM of H₂O₂ dissolved in PBS containing DCFH-DA was added to appropriate wells. The plate was incubated for another hour. Lastly, the cells were washed with 1X PBS and lysed using lysis buffer, and 100 µL was transferred from each well to a flat black clear bottom plate in triplicate. The plate was read at 480 nm/530 nm using a Spectramax i3 plate reader. The DMSO values were subtracted from each of the treated samples. The experiment was repeated three times with triplicate samples.

A Transwell membrane (Corning) with 8-µm pole size was used for Boyden chamber invasion assays. The membrane was coated with gelatin-coating solution containing 0.03% gelatin and 0.1% acetic acid for 3 h in a 37°C CO₂ incubator. The media, containing 1% or 3% FBS plus either DMSO or 25 µM riluzole was placed into lower chamber. Media containing 0.1% FBS was used as a negative control. Cells from OS cell lines and patient-derived lines (U2OS, MG63, MG63.3, SaOS2, LM7 HOS, HOS-MNNG,143B, OS24, OS29, OS33, and OS69) were harvested from the culture plate using trypsin and were resuspended at a density of 4–6 x 10⁵ cells/ml in culture medium containing 0.1% FBS. The media with cells were loaded into the upper chamber at 50 µl/well. This Transwell chamber was placed for 17–20 h at 37°C in a humidified 5% CO₂ incubator. The membrane with invaded cells was fixed in methanol for 1 h in a −20°C freezer. The invaded cells were then stained using 0.1% crystal in 20% methanol. The non-invaded cells on the upper chamber surface of the Transwell membrane were wiped using wipes soaked in 30% glycerol. The invaded cells were quantified using an optical microscope. The growth inhibition by 25 μM Riluzole was measured for MG63, MG63.3, HOS, MNNG and SaoS2 after 18 hours of treatment to ensure that the growth inhibition was not contributing to the inhibition of invasion.

OS cells (LM7, MG63, MG63.3, HOS, HOS-MNNG, 143B, OS24, OS29, OS33, and OS69) were treated with either DMSO or 25 µM riluzole in medium containing 0.1% FBS for 48 h. The supernatants were collected and centrifuged to remove cells. After 48 h, the harvested protein content in the conditioned media were quantified using a Bradford assay. Next, the media was mixed with 5X nonreducing buffer and loaded onto an 8% SDS-PAGE gel containing 0.1% gelatin. After electrophoresis, the gels were washed three times with a wash buffer (2.5% Triton X-100 in distilled water). The gels were incubated with a developing buffer containing 50 mM tris(hydroxymethyl)aminomethane and 5 mM CaCl₂ at 37°C for 48 h in a shaking incubator. Subsequently, the gels were stained with Coomassie blue staining buffer, and the unstained regions were quantified using the ImageJ program to examine MMP2 activity.

The mRNA and protein quantifications in the cell lines and PDX were analyzed by one-way analysis of variance (ANOVA), followed by Dunnett’s test. MTT and glutamate assays were independently repeated three times, and statistical significance was determined using one-way ANOVA followed by Dunnett’s post hoc analysis. ROS assays were performed using three biological replicates, and the results were statistically examined by one-way ANOVA followed by Dunnett’s post hoc analysis. Invasion assays and gelatin zymography were conducted with three biological replicates, and the results were statistically examined by t-tests. Graphs were created using Prism 9.

We selected 11 cell lines with low or high metastatic potential ( Table 1 ), hFOB, 8 human OS lines (SaOS2, LM7, U2OS, MG63, MG63.3, HOS, HOS-143B, HOS-MNNG), and 2 murine OS lines (OS482 and K7M2). The metastatic potential of these cell lines has been evaluated by other labs (31, 32). The lines varied in their mutational status with respect to p53, RB, CDKN2A, and Myc expression. Genetic analysis of the PDX cells using the MSK-IMPACT (Integrated Mutation Profiling of Actionable Cancer Targets) tumor-profiling test revealed mutations in p53 and/or amplifications in other genes ( Table 1 ). Specifically, OS24 carries a EIF5-TP53 fusion gene, an X361_splice in NCOR1 (nuclear receptor corepressor 1) that results in a truncated protein, a mutation in DOT1L (histone methyltransferase) at K97R, and a point mutation in FGFR3 (fibroblast growth factor receptor 3) at K501M. OS29 has amplifications of the genes for nibrin (NBN1), which is involved in DNA repair; cyclin D3 (CCND3); vascular endothelial growth factor A (VEGFA), which is involved in angiogenesis; and notch receptor 4 (NOTCH4). OS33 has a deletion of RASA1, a negative regulator of Ras activity which may lead to constitutive Ras activity in these cells. OS69 has a missense point mutation at R337C in p53, which is important for oligomerization of p53; this germline mutation is seen in Li Fraumeni and related syndromes, which predispose patients to developing OS (33, 34).

The clinicopathological characteristics of the patients from whom the PDX lines were derived are shown in Table 2 . OS24 was isolated from a lung metastasis, while OS29, OS33, and OS69 were isolated from primary bone tumors. Among the OS lines, hFOB had high expression of wild type p53 and HOS, MNNG, and 143B had high expression of mutated p53 (R156P), as previously shown (35), but as expected p53 was not expressed by MG63, MG63.3, SaOS2, or LM7 (28, 36) ( Figure 1A ). Among the PDX cells, OS24, OS33, and OS69 expressed p53, but OS29 did not express a full-length p53 protein ( Figure 1A ). Thus, the cell and PDX lines evaluated in the study carry a variety of mutations, providing the desired heterogeneity for testing the efficacy of riluzole as a treatment for OS.

Autocrine glutamate signaling may play an important role in the proliferation of OS, and two OS cell lines, LM7 and MG63, have been shown to secrete glutamate (23, 37). To determine if glutamate secretion is common among OS cells, we measured secreted glutamate over seven days in culture media from the primary human OS cell lines hFOB, MG63, HOS, SaOS2, and U2OS ( Figure 1B ), as well as from the murine line K7M2. The results showed a steady increase in glutamate secretion over seven days, with the highest secretion by K7M2 (0.37 mM) followed by HOS and hFOB. The human metastatic OS cell lines 143B, HOS-MNNG, LM7, and MG63.3, and the mouse metastatic cell line OS482, also showed steady increases in glutamate secretion over seven-day period ( Figure 1C ). Glutamate levels were higher in LM7 (0.624 mM), 143B (0.648 mM), MG63.3 (0.545 mM), and K7M2 (0.478 mM) than in hFOB (0.26 mM). Moreover, the nonmetastatic cell lines, such as MG63, and HOS, secreted less glutamate than the metastatic cell lines, such as LM7, MG63.3, and 143B. Among the PDX lines, glutamate levels in OS29 and OS33 on day 7 were higher to those of hFOB, while the glutamate levels in OS69 were comparable to hFOB and those in OS24 were lowest of all ( Figure 1D ). Since dead cells release glutamate into the media, which may interfere with measurement of glutamate secretion, we performed trypan blue staining and confirmed the viability of the tested cells (data not shown). Overall, our results demonstrated that the primary and metastatic OS cell lines and the PDX cells all secreted glutamate, and that levels were generally highest among the metastatic lines.

Our previous work demonstrated that the proliferation of LM7 depends on signaling through the mGluR5 glutamate receptors (23). Moreover, mGluR1 glutamate receptors have been shown to be involved in proliferation of melanoma cells (38). Therefore, we measured mRNA expression of mGluR1 and mGluR5 in human OS cell lines using qPCR, respectively ( Figures 1E, F ). Compared to hFOB, the cell lines MG63, MNNG, SaOS2, LM7, and U2OS had significantly higher expression of mGluR1 mRNA. Similarly, compared with hFOB, LM7 and U2OS cells had higher expression of mGluR5 mRNA, but SaOS2 (the parental cell line of LM7) did not. In essence, the expression of mGluR1 and mGluR5 differed between the primary parental cell lines and the derived metastatic cell lines.

Export of glutamate into the extracellular space occurs through two routes, one mediated by vesicular glutamate transport and the other through xCT (SLC7A11), a functional component of system Xc- (39). Vesicular glutamate transporter (VGLUT) plays an important role in storage and transport of L-glutamate via exocytosis (40). In osteoclasts, the vesicular transport was demonstrated by Morimoto et al. via the transcytosis of the vesicular glutamate transporter 1 (vGLUT1) (40). xCT exports glutamate and imports cystine, which is required for intracellular glutathione synthesis (41). Cancer cells secrete glutamate via xCT (42), and xCT overexpression contributes to proliferation, metastasis, and drug resistance (43, 44).

We measured the expression of vGLUT1 and xCT in all OS cell lines and PDXs to determine the route of extracellular glutamate in each. We found vGLUT1 protein expression in hFOB, MG63, MG63.3, HOS, SaOS2, LM7, and U2OS cells, but little or no expression in HOS-MNNG and 143B cells ( Figure 1G ). All four PDXs expressed vGLUT, and levels did not differ substantially among them. We observed xCT protein expression in hFOB, MG63, MG63.3, HOS, HOS-MNNG, 143B, SaOS2, LM7, and U2OS cells; among PDX cells, OS24 and especially OS29 expressed xCT ( Figure 1H ). Thus, we observed vGLUT and xCT expression in most of the human OS cell lines, and xCT expression in two of the four PDX cells.

Riluzole is widely used to block glutamate secretion for several neurodegenerative diseases in which excessive secretion is the underlying cause (45, 46). Similarly, riluzole has been tested in many types of cancer that depend on glutamate for growth and proliferation (13); for example, in OS, riluzole blocks proliferation in LM7 and OS482 cells. To investigate the drug’s effect on other OS cell lines, we performed a dose-response growth inhibition assay. We seeded the cells at concentrations designed to ensure that the control cells would not be dead at 72 h and used drug concentrations ranging from 5 µM to 100 µM. The IC₅₀ of riluzole varied among cell lines ( Figures 2A–K ). For all cell lines except LM7, there was significant cell death at 25 µM riluzole; for 143B, the IC₅₀ remained at 100 µM. We observed proliferation instead of inhibition of U2OS cells at 5 µM, which was consistent across independent experiments. Overall, the data clearly demonstrated that the OS cell lines showed significant, dose-dependent growth inhibition in response to riluzole.

We then tested the patient-derived lines. As Figures 2L–O show, the IC₅₀ was 10 µM for OS24 and OS29 and 25 µM for OS33. Although OS69 showed variable response at 10 µM, the growth inhibition at 25 µM and above was consistently over 80%. These results demonstrate that the PDX lines consistently respond to riluzole in a dose-dependent manner similar to that of the OS cell lines in vitro. The IC₅₀ values for riluzole are shown for all cell lines and PDX cells in Table 3 .

Riluzole inhibits proliferation in hepatocellular carcinoma by increasing ROS (25), and we demonstrated a similar ROS increase in LM7 cells (27). To test if this is a common response to riluzole in OS cell lines and patient-derived cells, we measured ROS production as an indicator of oxidative stress levels. We used riluzole at 50 μM and 100 μM to assess whether any increase in ROS was dose dependent. Of the 8 tested cell lines (hFOB, SaOS2, LM7, U2OS, MG63, MG63.3, HOS, and 143B), all but 143B showed a significant increase in ROS production in response to riluzole ( Figure 3 ). We also found significant increases in the four patient-derived lines, though intriguingly in OS69 the increase was apparent at 50 µM but not at 100 µM. Collectively, the data demonstrates that ROS production is a common response to riluzole in OS cell lines and our patient-derived lines.

N-Acetyl-L-Cysteine (NAC) is an established antioxidant agent that has been proved to be effective in compensating for the changes in the ROS (47). To test if cell growth inhibiting by riluzole is due to ROS increase, we performed growth inhibition in the presence of NAC in MG63.3 and LM7 cells. We found the growth inhibition by riluzole was significantly rescued by NAC ( Supplementary Figure S1 ).

One of the hallmarks of cancer cells is the ability to invade surrounding tissues (48). We previously used a scratch assay to demonstrate that riluzole inhibits migration of LM7 cells (23). Here, to determine the drug’s effects on invasion properties of OS cell lines and patient-derived lines, we measured cell invasion using a Boyden chamber assay in the presence or absence of riluzole. To prevent the interference of cell death in the invasion assay, we used riluzole at a lower dose (25 μM) and set the assay duration to 18 hours. We found significant inhibition of cell invasion in the OS cell lines MG63, MG63.3, SaOS2, LM7, U2OS, HOS, HOS-MNNG, and 143B ( Figures 4A–H ), and in the patient-derived lines OS24, OS29, OS33, and OS69 ( Figures 4I–L ).

Matrix metalloproteinases (MMPs), a family of extracellular matrix remodeling enzymes, play an important role in invasion and metastasis of cancer cells by cleaving components of the extracellular matrix (49). Specifically, MMP2 (gelatinase A) and MMP9 (gelatinase B) are associated with tumor spread and cell invasion (50, 51). Because riluzole inhibits the invasion of OS cells, we assessed activity of MMP2 and MMP9 to investigate their potential role in the invasive properties of OS cells. We performed gelatin zymography for OS cell lines and patient-derived cells in the presence or absence of riluzole. We found significant decrease in MMP2 activity in MG63, MG63.3, HOS, HOS-MNNG, and 143B, but not in LM7 cells or the four patient-derived lines ( Figure 5 ). Moreover, riluzole did not increase MMP9 activity in any of the OS cell lines ( Figure 5 ). Thus, it appears that riluzole inhibits cell invasion via inhibition of MMP2 activity in MG63, MG63.3, HOS, HOS-MNNG, and 143B but through other mechanisms in LM7, OS24, OS29, OS33, and OS69 cells.

To determine if riluzole regulates transcription of MMP2 and MMP9, we measured mRNA levels of these enzymes in the 6 OS and PDX lines. We found that the mRNA levels for MMP2 were significantly reduced in MG63, MG63.3, HOS, and HOS-MNNG cells, but not in LM7 or 143B ( Figure 6 ). On the contrary, MMP2 mRNA levels were significantly increased in OS33 and OS69. The only significant changes in MMP9 mRNA expression levels were upregulation in MG63.3 and OS69 ( Figure 6 ).