Novel Thieno [2,3-b]pyridine Anticancer Compound Lowers Cancer Stem Cell Fraction Inducing Shift of Lipid to Glucose Metabolism

Due to the role of cancer stem cells (CSCs) in tumor resistance and glycosphingolipid (GSL) involvement in tumor pathogenesis, we investigated the effect of a newly synthesized compound (3-amino-N-(3-chloro-2-methylphenyl)-5-oxo-5,6,7,8-tetrahydrothieno[2,3-b]quinoline-2-carboxamide 1 on the percentage of CSCs and the expression of six GSLs on CSCs and non-CSCs on breast cancer cell lines (MDA-MB-231 and MCF-7). We also investigated the effect of 1 on the metabolic profile of these cell lines. The MTT assay was used for cytotoxicity determination. Apoptosis and expression of GSLs were assessed by flow cytometry. A GC–MS-coupled system was used for the separation and identification of metabolites. Compound 1 was cytotoxic for both cell lines, and the majority of cells died by treatment-induced apoptosis. The percentage of CSCs was significantly lower in the MDA-MB-231 cell line. Treatment with 1 caused a decrease of CSC IV6Neu5Ac-nLc4Cer+ MDA-MB-231 cells. In the MCF-7 cell line, the percentage of GalNAc-GM1b+ CSCs was increased, while the expression of Gg3Cer was decreased in both CSC and non-CSC. Twenty-one metabolites were identified by metabolic profiling. The major impact of the treatment was in glycolysis/gluconeogenesis, pyruvate and inositol metabolism. Compound 1 exhibited higher potency in MBA-MB-231 cells, and it deserves further examination.


Introduction
Cancer is a leading cause of death in almost every country of the world. Among all cancers, breast cancer is present in 11.7% of new cases in both sexes, while in women the incidence is 24.5%, with a mortality rate of 15.5% [1].
Triple negative breast cancer, TNBC, is an aggressive and heterogeneous disease, defined by the lack of receptors for estrogen (ER), progesterone (PR) and receptor 2 for epidermal factor (HER2), and accounts for 10-25% of all breast cancers. Due to the lack of targeted molecular therapy, this type of cancer is the most aggressive, with the worst prognosis [2,3].
Cancer stem cells (CSC) are a small, proliferating subpopulation of cancer cells that have the ability to self-renew and reform a primary tumor [4].
Breast cancer stem cells (BCSCs), defined by the CD44 + /CD24 − phenotype, show unique characteristics, including tumorigenicity, self-renewal, differentiation, metastatic Metabolic profiling of cancer cells has a growing role in the elucidation of the molecular basis of cancer development, progression and prognosis. It represents a powerful approach for the characterization of phenotypes and the distinction of specific metabolic states due to environmental alterations. The early observations by Dr Otto Warburg revealed that fundamental metabolic differences exist between malignant tumor cells and adjacent normal cells [28]. The molecular phenotyping of different cancer cell lines and monitoring of their response to different drug treatments can help in the search of new candidate biomarkers and to elucidate new therapeutic targets. These observations suggest that targeting the peculiar metabolic pathways in cancer might be an effective strategy for cancer therapy [29].
In this study, we investigated the effect of a newly synthesized thieno [2,3-b]pyridine compound on the metabolic profile of two breast cancer cell lines with different hormone receptor expressions: MDA-MB-231 (triple negative) and MCF-7 (ER+, PR+, HER2-). A GC-MS coupled system was used for separation and identification of metabolites. , and cell metabolism was evaluated by the 3-(4,5-dimethylthiazolyl-2)-2,5-diphenyltetrazolium bromide (MTT) assay. Data are expressed as a mean from the experiment performed in triplicate ± SD. Columns, mean of viable cells; bars, SD (standard deviation); a, p < 0.05; b, p < 0.01; c, p < 0.001. SD, standard deviation.

Apoptosis Rate
To determine the type of cell death induced by 48 h treatment with 2 µM (the average IC50 for both cell lines) of compound 1, we determined early (Annexin-V + PI − subpopulation), late (Annexin-V + PI + subpopulation) and percentage of total apoptosis. The treatment induced an increase of the percentage of apoptotic cells in both cell lines, as shown in , and cell metabolism was evaluated by the 3-(4,5-dimethylthiazolyl-2)-2,5-diphenyltetrazolium bromide (MTT) assay. Data are expressed as a mean from the experiment performed in triplicate ± SD. Columns, mean of viable cells; bars, SD (standard deviation); a, p < 0.05; b, p < 0.01; c, p < 0.001. SD, standard deviation.
The inhibitory concentration for 50% of cells (IC 50 ) for the MDA-MB-231 cell line after 48 h of treatment was 2.082 µM, whilst for the MCF7 cell line it was 2.053 µM.

Apoptosis Rate
To determine the type of cell death induced by 48 h treatment with 2 µM (the average IC 50 for both cell lines) of compound 1, we determined early (Annexin-V + PI − subpopulation), late (Annexin-V + PI + subpopulation) and percentage of total apoptosis. The treatment induced an increase of the percentage of apoptotic cells in both cell lines, as shown in Figure 2. In the MDA-MB-231 cell line, compound 1-treated cells showed a significant increase in early (A), late (B) and total (C) apoptosis, while in the MCF-7 cell line, the increase was found in late (F) and total (G) apoptosis compared with non-treated cells.

Mammosphere Formation
To determine whether the cancer stem cells are sensitive to compound 1, the number of mammospheres formed after 7 days of treatment was counted ( Figure 3A,B). After treatment with compound 1, the number of mammospheres was reduced in both cell lines, but significantly only in MCF-7 ( Figure 3B). However, the effect of compound 1 was observed in the size and shape of the mammospheres-in both cell lines mammospheres formed in wells with the compound 1 were significantly smaller with an irregular shape ( Figure 3C,D).

Cancer Stem Cells
To investigate whether compound 1 contributes to the reduction of the cancer stem cells (CSC) number, we determined CD44 + /CD24 − subpopulation in both the cell lines. Treatment with 1 resulted in a statistically significant decrease of the cancer stem cells in the MDA-MB-231 cell line, known to have a higher CSC population than other breast cancer cell lines ( Figure 4A). The number of CSC in the MCF7 cell line also decreased after treatment with 1, but not in a statistically significant manner ( Figure 4B). , and cell metabolism was evaluated by the 3-(4,5-dimethylthiazolyl-2)-2,5-diphenyltetrazolium bromide (MTT) assay. Data are expressed as a mean from the experiment performed in triplicate ± SD. Columns, mean of viable cells; bars, SD (standard deviation); a, p < 0.05; b, p < 0.01; c, p < 0.001. SD, standard deviation.

Apoptosis Rate
To determine the type of cell death induced by 48 h treatment with 2 µM (the average IC50 for both cell lines) of compound 1, we determined early (Annexin-V + PI − subpopulation), late (Annexin-V + PI + subpopulation) and percentage of total apoptosis. The treatment induced an increase of the percentage of apoptotic cells in both cell lines, as shown in Figure 2. In the MDA-MB-231 cell line, compound 1-treated cells showed a significant increase in early (A), late (B) and total (C) apoptosis, while in the MCF-7 cell line, the increase was found in late (F) and total (G) apoptosis compared with non-treated cells.  To determine whether the cancer stem cells are sensitive to compound 1, the number of mammospheres formed after 7 days of treatment was counted ( Figure 3A,B). After treatment with compound 1, the number of mammospheres was reduced in both cell lines, but significantly only in MCF-7 ( Figure 3B). However, the effect of compound 1 was observed in the size and shape of the mammospheres-in both cell lines mammospheres formed in wells with the compound 1 were significantly smaller with an irregular shape ( Figure 3C,D).

Cancer Stem Cells
To investigate whether compound 1 contributes to the reduction of the cancer stem cells (CSC) number, we determined CD44 + /CD24 − subpopulation in both the cell lines. Treatment with 1 resulted in a statistically significant decrease of the cancer stem cells in the MDA-MB-231 cell line, known to have a higher CSC population than other breast cancer cell lines ( Figure 4A). The number of CSC in the MCF7 cell line also decreased after treatment with 1, but not in a statistically significant manner ( Figure 4B).

Expression of Terminally Sialylated Gangliosides on CSCs and Non-CSCs
Sialylation is involved in cell fate processes during development, reprogramming and tumor progression [30,31]. Therefore, the expression of terminally sialylated gangliosides at CSCs (CD44 + /CD24 − subpopulation) and non-CSCs (CD44 + /CD24 + , CD44 − /CD24 − and CD44 − /CD24 + subpopulation; abbreviated as non-CD44 + /CD24 − ) was studied, with the aim of determining whether the cytotoxic effects of 1 are mediated via different GSL membrane contents. The expression of each GSL per one cell is represented with the geometric mean fluorescence intensity (GMI). The portion of the cells that are GSL positive is an

Expression of Terminally Sialylated Gangliosides on CSCs and Non-CSCs
Sialylation is involved in cell fate processes during development, reprogramming and tumor progression [30,31]. Therefore, the expression of terminally sialylated gangliosides at CSCs (CD44 + /CD24 − subpopulation) and non-CSCs (CD44 + /CD24 + , CD44 − /CD24 − and CD44 − /CD24 + subpopulation; abbreviated as non-CD44 + /CD24 − ) was studied, with the aim of determining whether the cytotoxic effects of 1 are mediated via different GSL membrane contents. The expression of each GSL per one cell is represented with the geometric mean fluorescence intensity (GMI). The portion of the cells that are GSL positive is an interesting parameter, but it is of less impact in comparison to GMI. Treatment with 1 resulted only in a significant percentage decrease of CSC IV 6 Neu5Ac-nLc 4 Cer-positive cells in the MDA-MB-231 cell line ( Figure 5A). Expression of IV 3 Neu5Ac-nLc 4 Cer was not affected by treatment with compound 1 in both cell lines, as shown in Figure 5.
with compound 1 for 48 h in MDA-MB 231 (A) and in the MCF-7 cell line (B). Data represented are expressed as a mean from experiment performed in triplicate ± SD. Columns, mean of cells; bars, SD; *** p < 0.001. CSCs cancer stem cells; SD, standard deviation.

Expression of Terminally Sialylated Gangliosides on CSCs and Non-CSCs
Sialylation is involved in cell fate processes during development, reprogramming and tumor progression [30,31]. Therefore, the expression of terminally sialylated gangliosides at CSCs (CD44 + /CD24 − subpopulation) and non-CSCs (CD44 + /CD24 + , CD44 − /CD24 − and CD44 − /CD24 + subpopulation; abbreviated as non-CD44 + /CD24 − ) was studied, with the aim of determining whether the cytotoxic effects of 1 are mediated via different GSL membrane contents. The expression of each GSL per one cell is represented with the geometric mean fluorescence intensity (GMI). The portion of the cells that are GSL positive is an interesting parameter, but it is of less impact in comparison to GMI. Treatment with 1 resulted only in a significant percentage decrease of CSC IV 6 Neu5Ac-nLc4Cer-positive cells in the MDA-MB-231 cell line ( Figure 5A). Expression of IV 3 Neu5Ac-nLc4Cer was not affected by treatment with compound 1 in both cell lines, as shown in Figure 5. after treatment with compound 1 in duration of 48 h. Data are expressed as a mean from experiment performed in triplicate ± SD. Columns, mean of viable cells; bars, SD; * p < 0.05. CSC, CD44 + /CD24 − cells; non-CSC, CD44 + /CD24 + , CD44 − /CD24 − and CD44 − /CD24 + cells; GMI, geometric mean fluorescence intensity; SD, standard deviation. The designation of the gangliosides follows the IUPAC-IUB recommendations [32] and the nomenclature of Svennerholm [33]: Neu5Ac, N-acetylneuraminic acid; nLc4Cer, neolactotetraosylceramide; IV 3 Neu5Ac-nLc4Cer; IV 6 Neu5Ac-nLc4Cer.

Expression of Gangliosides and Neutral GSLs with Terminal GalNAc Residue on CSCs and Non-CSCs
The expression of these glycosphingolipids was not affected by treatment with compound 1 in the MDA-MB-231 cell line ( Figure 6, upper row). In the MCF-7 cell line, the percentage of GalNAc-GM1b-positive cells was significantly increased within the CSC cell subpopulation (CD44 + /CD24 − ), whilst the percentage of Gg 3 Cer-positive cells was decreased within both subpopulations, CSC and non-CSC ( Figure 6, lower row).

Comparative Metabolic Profiling of Breast CSC Using GC-MS
The aim of this study was to compare the metabolic response of MDA-MB-231 and MCF-7 breast cancer cells induced by compound 1. Metabolic profiling was performed

Comparative Metabolic Profiling of Breast CSC Using GC-MS
The aim of this study was to compare the metabolic response of MDA-MB-231 and MCF-7 breast cancer cells induced by compound 1. Metabolic profiling was performed using GC-MS, and 21 metabolites were identified (Table 1). Only compounds listed in the Human Metabolome Database (HMDB 4.0) were selected. Ultimately, entries with more than 50% of missing values were excluded from the study. The objective was to identify the effect of compound 1 on each metabolite and to find metabolites that were significantly different between treated and control cells. The test of statistical significance revealed that the results for MDA-MB-231 cells are more significant than for MCF-7 cells. As listed in Table 1, 6 metabolites were significantly different (p < 0.05) from the control in MDA-MB-231 cells and just one metabolite in MCF-7 cells. One-way analysis of variance (ANOVA) was used for statistical significance among MDA-MB-231 and MCF-7 cells. Three significant features (pyruvate, myo-inositol and lactate) were found using the given threshold (p < 0.05) for the one-way ANOVA and post hoc Tukey test. PCA results showed that compound 1 treatment causes metabolic alternations in both cell lines. PCA score plots (Figure 7) visualize the clustering of treated group versus control. PC1 accounts for 67% and 50% of the variation in the data in MDA-MB-231 and MCF-7 line, respectively. The score plots showed that the control group and treated group could be clearly distinguished. Quantitative enrichment analysis was employed to identify patterns of metabolite concentrations and to help elucidate possible biological mechanisms. Results were considered statistically significant with p < 0.05, and no statistical difference was established for MCF-7 cells. The results of the analysis indicate that treatment had a major impact in glucose/energy metabolism, specifically in glycolysis/gluconeogenesis, the pyruvate metabolism, Warburg effect and inositol metabolism (Figure 8).   Quantitative enrichment analysis was employed to identify patterns of metabolite concentrations and to help elucidate possible biological mechanisms. Results were considered statistically significant with p < 0.05, and no statistical difference was established for MCF-7 cells. The results of the analysis indicate that treatment had a major impact in glucose/energy metabolism, specifically in glycolysis/gluconeogenesis, the pyruvate metabolism, Warburg effect and inositol metabolism ( Figure 8). Quantitative enrichment analysis was employed to identify patterns of metabolite concentrations and to help elucidate possible biological mechanisms. Results were considered statistically significant with p < 0.05, and no statistical difference was established for MCF-7 cells. The results of the analysis indicate that treatment had a major impact in glucose/energy metabolism, specifically in glycolysis/gluconeogenesis, the pyruvate metabolism, Warburg effect and inositol metabolism ( Figure 8).

Discussion
It is well known that cancer is linked to other chronic medical conditions [34]. For example, patients diagnosed with metabolic syndrome show an increased incidence and aggressiveness of tumor formation [35]. In this study, metabolic profiling of two different breast cancer cell lines revealed that compound 1 increased pyruvate and decreased lactate, phosphate, myo-inositol, glycerol monopalmitate and glycerol monostearate in MDA-MB-231 cells. An inverse effect was noticed for two metabolites in MCF-7 cells: increased lactate and glyceryl-monostearate (Figure 9). Compound 1 treatment did not influence glycosphingolipid expression per one cell, but the percentage of IV 6 NeuAc-nLc4-positive cells was decreased in MDA-MB-231 CSC, while MCF-7 CSC involved increased GalNAc-GM1b+ and decreased Gg3Cer+ subpopulation percentage. Gg3Cer+ subpopulation was also decreased within the non-CSC of the MCF-7 cell line, which is in accordance with finding of Liang et al. After the knockdown of ST8SIA1 and B4GALNT1, they found significantly reduced expression of GD2 and GD3 gangliosides and phenotype change from CSC to a non-CSC [16]. Namely, Gg3Cer-neutral GSL is also synthesized with B4GALNT1 ( Figure 10). Terminally α2-6-sialylated gangliosides of the neolactose series IV 6 NeuAc-nLc4 are recognized earlier as receptors for rViscumin, which is used as an adjuvant cancer therapy [41]. Recently, we described increased IV 6 NeuAc-nLc4 expression in both CSCs and non-CSCs of the MDA-MB-231 cell line after treatment with a different compound that belongs also to the thieno [2,3-b]pyridine group of molecules [27]. Gg3Cer has been detected in murine lymphosarcoma cell lines [42] and in myeloid leukemia cells as major neutral glycosphingolipid [43].
Compound 1 was more cytotoxic to MDA-MB-231 cells, and the results showed that it shifted glycolysis to the aerobic pathway (increased pyruvate, decreased lactate), as seen by the significant reduction in the number of cancer stem cells, which prefer anaerobic glycolysis. The opposite could be observed in MCF-7 cells, where the compound increased lactate, indicating a shift to anaerobic glycolysis. The decrease of phosphate after compound 1 treatment in MDA-MB-231 cells could be caused by the increased binding of phosphate ions to their transporters in the inner mitochondrial membrane during aerobic glycolysis. The binding of phosphates to its transporters is favored with hydrogen ions present in intermembrane mitochondrial space during aerobic glycolysis.
Inverse compound 1 effect upon the glycerol-monostearate concentration in MDA-MB-231 in relation to MCF-7 cells could be developed due to different growth factor receptors, present in two cell lines. Genes involved in lipid metabolism are highly enriched in diverse models of cellular transformation and are thereby linked to cancer [34]. These genes are for OLR1, oxidized low-density lipoprotein receptor 1; SNAP23, crucial protein for storage of neutral lipids in lipid droplets and translocation of GLUT4 to the plasma membrane [36]; VAMP4, protein-driving lipid droplet forming, together with SNAP23; SCD1, stearoyl-CoA desaturase-1; and SREBP1, master lipogenesis regulator, sterol regulatory element binding protein 1. SREBP1 is targeted to inhibit lipid metabolism, thereby obstructing epithelialmesenchymal transition in breast cancer [37]. Fatty acid synthase plays an important role in cancer pathogenesis [38]. Growth factor-dependent or independent autophosphorylation of receptor tyrosine kinases leads to signal transduction cascades. MCF-7 cells express estrogen, progesterone and androgen receptors while MDA-MD-231 cells express only androgen receptors [39]. However, progesterone can transmit its signal in MDA-MB-231 cells by binding to progesterone receptor membrane component 1, which has been shown to regulate some cancer hallmarks [40]. Steroid hormones including estradiol, progestins and androgens trigger transduction mechanisms that ultimately stimulate fatty acid synthase expression through the modulation of the transcription factor sterol regulatory elementbinding protein 1c (SREBP1c), which binds to and activates sterol regulatory elements in the promoter region of fatty acid synthase [38]. The lowering of glycerol monopalmitate and glycerol monostearate in MDA-MB-231 cells by compound 1 indicates its influence upon the reversal of lipid to glucose metabolism. This shift includes the inhibition of the pentose phosphate pathway, which is the source of reductive power (NADPH) for palmitate and stearate synthesis, and the inhibition of citrate transport from mitochondria in the cytosol [38]. Fatty acid synthesis requires cleavage of the citrate in cytosol to yield acetyl-CoA, the precursor of palmitate and stearate.
Compound 1 treatment did not influence glycosphingolipid expression per one cell, but the percentage of IV 6 NeuAc-nLc 4 -positive cells was decreased in MDA-MB-231 CSC, while MCF-7 CSC involved increased GalNAc-GM1b+ and decreased Gg3Cer+ subpopulation percentage. Gg3Cer+ subpopulation was also decreased within the non-CSC of the MCF-7 cell line, which is in accordance with finding of Liang et al. After the knockdown of ST8SIA1 and B4GALNT1, they found significantly reduced expression of GD2 and GD3 gangliosides and phenotype change from CSC to a non-CSC [16]. Namely, Gg3Cer-neutral GSL is also synthesized with B4GALNT1 ( Figure 10). Terminally α2-6-sialylated gangliosides of the neolactose series IV 6 NeuAc-nLc 4 are recognized earlier as receptors for rViscumin, which is used as an adjuvant cancer therapy [41]. Recently, we described increased IV 6 NeuAc-nLc 4 expression in both CSCs and non-CSCs of the MDA-MB-231 cell line after treatment with a different compound that belongs also to the thieno pyridine group of molecules [27]. Gg3Cer has been detected in murine lymphosarcoma cell lines [42] and in myeloid leukemia cells as major neutral glycosphingolipid [43].  [32]. Only transferases B4GALNT1, B4GALT5 and ST3GAL2, which are among the most consistently malignancy-associated enzymes [44], are marked at corresponding steps of synthesis. The relative concentrations of the metabolites.
The gene-coding GalNAc-transferase (GALNT2-SRD5A3) that binds glycosyl residues to glycoproteins belongs to a family of genetic alteration-driven genes and drug target genes acting as prognostic biomarkers in breast cancer [45]. GALNT2 gene-coding O-GalNAC transferase 2, which synthesizes Tn-antigen in core O-glycosylation (mucin type) glycoproteins, is associated with high cholesterol and triglyceride levels [46]. Investigation by Pucci et al. point to GALNT2 as one of the most consistent malignancy-associated  [32]. Only transferases B4GALNT1, B4GALT5 and ST3GAL2, which are among the most consistently malignancy-associated enzymes [44], are marked at corresponding steps of synthesis. The relative concentrations of the metabolites.
The finding of an increased GalNAc-GM1b+ and decreased Gg3Cer+ subpopulation percentage in MCF-7 CSC indicates that the addition of neuraminic acid (Neu5Ac) and two sugar residues (GalNAcβ1-4Galβ1) to Gg3Cer favors the survival of MCF-7 CSC ( Figure 10).
The gene-coding GalNAc-transferase (GALNT2-SRD5A3) that binds glycosyl residues to glycoproteins belongs to a family of genetic alteration-driven genes and drug target genes acting as prognostic biomarkers in breast cancer [45]. GALNT2 gene-coding O-GalNAC transferase 2, which synthesizes Tn-antigen in core O-glycosylation (mucin type) glycoproteins, is associated with high cholesterol and triglyceride levels [46]. Investigation by Pucci et al. point to GALNT2 as one of the most consistent malignancy-associated enzymes together with B4GALNT1, B4GALT5, B3GNT5 and ST3GAL2 [44]. B4GALT5 and B4GALNT1 are involved in steps of Gg3Cer (called also asialo GM2) and GalNAc-GM1b synthesis, as shown in Figure 10.
GalNAc-GM1b is a ganglioside, acidic glycosphingolipid that contains sialic acid. ST3GAL2 catalyzes sialic acid transfer to Gg4Cer (called also asialo GM1). GalNAc-GM1b is described as minor in four human uterine cervical carcinoma-derived cell lines [47]. Sphingosine and ceramide ( Figure 10) can be composed of different fatty acids-for example, stearic acid-with 18 carbon atoms [48], but mostly from palmitic acid (C16), which is the major fatty acid present in metabolism [49]. Recently, Cruz et al. showed beneficial effects of ceramide use in nanomedicine, targeting it in combination with doxorubicin, within liposomes, to MDA-MB-231 cells [50]. MCF cells are able to escape from the ceramiderelated apoptosis by the activation of ganglioside synthesis [51]. In our study, compound 1 significantly increased early and late apoptosis in MDA-MB-231 cells, but only late apoptosis in MCF-7 cells. Only MCF-7 CSC showed an increased percentage of ganglioside GalNac-GM1b+ cells. We can assume that those cells contained lower ceramide levels and are thereby less sensitive to early apoptosis. Again, as well as for the aforementioned Gg3Cer, there is accordance of the GalNac-GM1b finding with the results of Liang et al. [16]. They found an increased level of mRNA coding ST3GAL2 glycosyltransferase, involved in GalNac-GM1b synthesis, in CSCs compared to non-stem breast cancer cells ( Figure 10).
In conclusion, compound 1 exhibited higher potency in MBA-MB-231 cells, the triple negative breast cancer cell line, and it deserves further examinations in in vitro and in vivo models.

Cytotoxic Activity Assay
To estimate cell viability, MTT was assessed [52]. Equal numbers of cells were seeded in three replicates. After overnight incubation, cells were treated with complete media or individual solutions of 1 at 0.05, 0.1, 0.25, 0.5, 1, 2.5, 5, 7.5, 10 and 25 µM in complete media, in three repetitions, for 4, 24, 48 and 72 h. Following treatment, cells were incubated with combined quenching and extraction step was performed, without cell scraping. 20 µL of ribitol (Sigma Aldrich, Steinheim, Germany) was added as an internal standard, and the sample was dried in a vacuum rotational evaporator (Laborota 4000-efficient, Heidolph, Schwabach, Germany).

Sample Derivatization and GC-MS Analysis
The derivatization process involved addition of 25 µL solution consisting of 20 mg/mL methoxyamine hydrochloride in pyridine, followed by constant shaking for 60 min at 50 • C and the addition of MSTFA (N-Methyl-N-(trimethylsilyl)trifluoroacetamide, Sigma Aldrich, Steinheim, Germany) + 1% TMCS (Trimethylchlorosilane, Sigma Aldrich, Steinheim, Germany), with incubation at 50 • C for 30 min for complete derivatization. The sample was dissolved in 100 µL pyridine.
The samples were analyzed using an Agilent 8890 GC system coupled with triple quad spectrometer system MS 7000D GC/TQ. The column was HP-5 MS (30m × 0.25 mm × 0.25 µm, Agilent, Santa Clara, CA, USA) with an oven program set on 60 • C maintained for 2 min, then increased to 210 • C at rate of 10 • C/min, ramped to 240 • C at rate of 5 • C/min and then ramped to 315 • C at a rate of 25 • C/min, and then held at 315 • C for 3 min.

GC-MS Data Preprocessing and Statistical Analysis
Agilent MassHunter Qualitative Analysis software was used for spectral processing (including peak picking, alignment, annotation and integration). Metabolites were identified using the NIST library. The intensity value for each metabolite was normalized to ribitol internal standard signal. A total of 21 metabolites were identified in cell culture samples.
MetaboAnalyst, a platform dedicated to metabolomics data analysis, was used to establish a panel of differentially expressed metabolites across these two cell lines.
Statistical significance analysis was performed using t-test. One-way ANOVA was used to compare the effect of 1 on both cell lines. Principal component analysis (PCA) was used as an unsupervised clustering method. Metabolic set enrichment analysis (MSEA) was employed to establish a link between metabolic fingerprints and metabolite concentration changes.
The IC 50 values were calculated using GraphPad Prism 7.0 (San Diego, CA, USA). For statistical analyses of apoptosis and glycosphingolipids expression, t-test with unequal variances, one-way ANOVA followed by post hoc Tukey test or Kruskal-Wallis, followed by Dunn's post hoc test, were performed using statistical software GraphPad Prism 7.0 (San Diego, CA, USA), with the significance set at p < 0.05.  Data Availability Statement: The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.