Purpose:

Metabolome analysis is an emerging method that provides insight into intracellular and physiologic responses. Methotrexate (MTX) is an antifolate that suppresses DNA syntheses by inhibiting dihydrofolate reductase. High-dose methotrexate treatment with deferred radiotherapy is a standard protocol in primary central nervous system lymphoma (PCNSL) treatments. However, most cases come to relapse-acquired resistance, in which the role of metabolic pathways is largely unknown.

Experimental Design:

Metabolome analysis in methotrexate-resistant PCNSL-derived cells (designated as TK-MTX and HKBML-MTX) was performed to detect alternative metabolites and pathways.

Results:

The metabolomic analyses using capillary electrophoresis-time-of-flight mass spectrometry detected 188 and 169 peaks in TK- and HKBML-derived cells, respectively, including suppression of central carbon metabolism, lipid metabolism, nucleic acid metabolism, urea cycle, branched chain and aromatic amino acids, and coenzyme metabolism. Particularly, whole suppressive metabolic pathways were demonstrated in TK-MTX, whereas HKBML-MTX indicated partially enhanced pathways of the urea cycle, amino acid metabolism, and coenzyme metabolism. Reciprocally detected metabolites for glycolysis, including induced glucose and reduced glycogen, and induced lactate and reduced pyruvate, in addition to increased lactate dehydrogenase activity, which is involved in Warburg effect. Thereby, ATP was increased in both methotrexate-resistant PCNSL-derived cells. Furthermore, we specifically found that PI3K/AKT/mTOR and RAS/MAPK signaling pathways were activated in TK-MTX but not in HKBML-MTX by growth rate with inhibitors and gene expression analysis, suggestive of cell type–specific methotrexate-resistant metabolic pathways.

Conclusions:

These results can help us understand targeted therapies with selective anticancer drugs in recurrent CNS lymphoma–acquired resistance against methotrexate.

Translational Relevance

Methotrexate (MTX) is an antifolate that suppresses DNA syntheses by inhibiting dihydrofolate reductase. High-dose methotrexate treatment with deferred radiotherapy is a first-line treatment in primary central nervous system lymphoma (PCNSL) treatments, whereas the median overall survival of patients with PCNSL shows a poorer prognosis and most cases come to relapse-acquired resistance to methotrexate. Here, we generated the two independent methotrexate-resistant PCNSL-derived cells, TK-MTX and HKBML-MTX, and demonstrated that TK-MTX and HKBML-MTX indicated whole suppressive metabolic pathways and partially enhanced pathways, respectively. Furthermore, we clarified enhanced glycolysis pathways including altered metabolites involved in the Warburg effect in both methotrexate-resistant PCNSL-derived cells with different manners. PI3K/AKT/mTOR and RAS/MAPK pathways were activated in TK-MTX but not in HKBML-MTX by growth rate with inhibitors and gene expression analysis, suggestive of cell type–specific methotrexate-resistant metabolic pathways. Hence, the results would help us understand targeted therapies with selective anticancer drugs in recurrent CNS lymphoma–acquired resistance against methotrexate.

Primary central nervous system (CNS) lymphoma (PCNSL) is an aggressive CNS tumor that is a rare type of extranodal non-Hodgkin lymphoma (NHL) including diffuse large B-cell lymphoma (DLBCL) localized into the brain, eye, spinal cord, and meninges, which is distinct from systemic DLBCL (1, 2). Most PCNSLs are immune privileged site–associated DLBCLs according to the WHO diagnostic criteria (1). PCNSLs account for 3% of all primary CNS tumors and 1% of NHLs in adults (3). Methotrexate is an antifolate that inhibits dihydrofolate reductase (DHFR) activity in purine and thymidine syntheses (4, 5), which controls the expression of glucocorticoid receptors in blood cells (6). High-dose methotrexate (HD-MTX) is used as a first-line treatment in PCNSL (7). Despite intensive treatments with the combinatorial use of HD-MTX–containing polychemotherapy regimens, including cyclophosphamide, pirarubicin, etoposide, vincristine, procarbazine with or without rituximab (8), and deferred whole brain radiotherapy, the median overall survival (OS) of patients with PCNSL associated with poorer prognosis is approximately 4 years (9, 10). Moreover, second-line treatments are also required for 10%–35% of patients with refractory diseases and for another 35%–60% or more who have relapse-acquired resistances (11).

The metabolism of cancer cells differs from that of most normal cells (12, 13). Cancer cells require altered metabolomic pathways to incorporate nutrients efficiently for immortal proliferation (14, 15) and for survival in different metabolic environments outside of normal tissues (16, 17). There are a few studies on metabolome analyses in NHL and DLBCL as revealed in searches of the Human Metabolome Database (HMDB). For example, electrospray ionization-liquid chromatography-tandem mass spectrometer (ESI-MS/MS) analyses in urine from 30 patients with NHL identified a low-mass ion pattern and decreased hypoxanthine (18). A recent metabolomic profiling demonstrated that panobinostat, a modified histone deacetylase inhibitor, alters lipid metabolism and the downstream survival signaling in DLBCL (19). In particular, panobinostat induces metabolic adaptations resulting in de novo pathways for choline and PI3K activation (19). However, to date, few studies have reported on the metabolomic analyses in CNS lymphoma with relapse-acquired resistance against methotrexate.

Here, we generated two independent methotrexate-resistant PCNSL-derived cell lines to conduct metabolome analyses and evaluate the differences between their metabolomic pathways and those of both original cell lines. Especially, we focused on glycolysis with lactate dehydrogenase (LDH) activity, ATP synthesis, glutamate metabolism, and mitochondrial activity. We also examined the gene expression for PI3K/AKT/mTOR signaling upstream of the glycolysis pathway, and defined cell type–specific characteristics of the two PCNSL-derived cells.

Cells

Primary central nervous system lymphoma (PCNSL) cell lines TK and HKBML were purchased from JCRB Cell Bank [National Institutes of Biomedical Innovation, Health and Nutrition (NIBIOHN), Osaka, Japan] on November 10, 2015, and RIKEN Cell Bank (RIKEN BRC, RIKEN BioResource Center, Tsukuba, Japan) on March 10, 2015, respectively, and were grown in RPMI1640 medium with 10% FBS and Ham's F-12 medium (Nacalai Tesque Inc.) with 15% FBS (Thermo Fisher Scientific), respectively, in 5% CO2 at 37°C. methotrexate-resistant cells were generated as described previously (20, 21). Briefly, TK cells (passage 23) were cultured with 1.0 × 10−6 mol/L methotrexate for 6 weeks following preculture with lower concentrations of methotrexate for 9 weeks. HKBML cells (passage 24) were cultured with 1.0 × 10−7 mol/L methotrexate for 6 weeks following preculture with lower concentrations of methotrexate for 4 weeks. TK and HKBML cells used in all experiments were cultured with 10% and 15% serum, respectively, in each medium. Mycoplasma contamination test was performed using e-Myco Mycoplasma PCR Detection Kit (iNtRON Biotechnology, Inc.) on February 20, 2020, which detected no contamination in the cells used.

Metabolome analysis

Metabolic extracts were prepared from 4–5 × 106 cells of TK-derived cells and 3–5 × 106 cells of HKBML-derived cells with methanol containing an internal standard solution (Human Metabolome Technologies, Inc.) and analyzed using a capillary electrophoresis (CE)-connected electrospray ionization-time-of-flight mass spectrometry (ESI-TOFMS) and CE-MS/MS system, according to the manufacturer's instruction (Human Metabolome Technologies, Inc.). Briefly, after removal of culture medium, cells were washed and then collected, according to the protocol E-140014 (Human Metabolome Technologies, Inc.). Cationic and anionic compounds were analyzed with the cation and anion mode of CE-TOFMS system with fused capillary inner diameter (i.d.) 50 μm × 80 cm (Agilent Technologies), respectively. Peaks detected by CE-TOFMS were extracted to obtain peak information, including mass-to-charge ratio (m/z), migration time (MT), and peak area (S/N > 3) using automatic integration software MasterHands (Keio University, Tsuruoka, Japan). Condition of measurements was as follows: (cation mode) sample injection; pressure injection 50 mbar, 10 seconds, CE voltage; positive, 27 kV, MS ionization; ESI Positive, MS capillary voltage; 4,000 V, MS scan range; m/z 50–1,000, sheath liquid; HMT Sheath Liquid (p/n: H3301-1020), and (Anion mode) sample injection; pressure injection 50 mbar, 25 seconds, CE voltage; positive, 30 kV, MS ionization; ESI Negative, MS capillary voltage; 3,500 V, MS scan range; m/z 50—1,000, sheath liquid; HMT Sheath Liquid (p/n: H3301-1020). The peaks were annotated by the HMT metabolite database (Human Metabolome Technologies, Inc.), based on their MTs in CE and m/z values determined by TOFMS. The tolerance annotation range of peaks was referred as ±0.5 min for MT and ±10 ppm for m/z. Concentrations of metabolites were estimated with normalization of the peak area of each metabolite (100 μmol/L) with respect to the area of the internal standard (200 μmol/L) and standard curves, which were detected by a one-point calibration. Relative area was calculated with a following formula: (target peak area)/(internal control area)/(sample volume). A metabolic pathway map was drawn with a Visualization and Analysis of Networks containing Experimental Data (http://vanted.ipk-gatersleben.de/). Triplicate samples were examined and relative areas of peaks were compared among groups with Welch t test (P < 0.05).

Cell proliferation assay

Cell proliferation assay was performed as described previously (22). Briefly, cells were seeded at 1 × 104 cells per well in 96-well plates. Cell proliferation was measured using a Cell Count Reagent SF, according to the manufacturer's instruction (Nacalai Tesque Inc.). Briefly, 10 μL of the Reagent SF solution was added to each well and incubated for 30 minutes in a 5% CO2 incubator at 37°C. The absorbance at 450 nm was measured with SpectraMax M2e (Molecular Devices, Tokyo, Japan).

Glycolysis, ATP synthesis, and glutamate metabolism

For glycolysis analysis, the amounts of glucose, glycogen, pyruvate, and lactate, and LDH activity were measured by colorimetric and fluorometric detection using Glucose Assay Kit (Ex/Em = 535/587 nm; Abcam), Glycogen Assay Kit (Ex/Em = 535/587 nm; Abcam), Pyruvate Assay Kit (Ex/Em = 535/587 nm; BioVision, Inc.), Lactate Assay Kit-WST (λ = 450 nm; Dojindo), and Cytotoxicity LDH Assay Kit-WST (λ = 490 nm; Dojindo), respectively, in accordance with the manufacturer's instructions. ATP synthesis was measured by fluorometric detection using the ATP Assay Kit (Ex/Em = 535/587 nm) (Abcam), according to the manufacturer's instruction. For glutamate metabolism, the amounts of glutamate and α-ketoglutamate, and glutamate dehydrogenase (GDH) activity were measured by colorimetric and fluorometric detection using Glutamate Assay Kit (Ex/Em = 540/590 nm) (Abcam), Alpha Ketoglutarate (alpha KG) Assay Kit (Ex/Em = 535/587 nm; Abcam), and Glutamate Dehydrogenase Activity Assay Kit (λ = 450 nm) (Abcam), respectively, according to the manufacturer's instructions. The absorbance and fluorescence signals were measured with SpectraMax M2e (Molecular Devices).

Mitochondrial activity assay

MitoBright Red (Ex/Em = 561/617 nm; Dojindo) and Mito-FerroGreen (Ex/Em = 488/550 nm; Dojindo) were used to evaluate activities of the mitochondrial membrane potential and the amount of Fe(II) oxide at Fe/S clusters in the mitochondria, respectively, according to the manufacturer's instructions. The fluorescence signals were measured with SpectraMax M2e (Molecular Devices).

Clustering analysis

Altered amounts of metabolites in the PCNSL-derived cells were clustered with the hierarchical method using the JMP built-in modules (SAS Institute, Inc.), as described previously (23).

Principal component analysis

Principal component analysis (PCA) was used to classify the metabolites detected by metabolomic analyses into the subgroups. PCA was performed using normalized values of metabolites using JMP built-in modules (SAS Institute Inc.), as described previously (8, 24).

In vivo tumorigenicity assay

The methotrexate-resistant cells and the parent cells were transplanted into right and left subcutaneous tissues in the back of nude mice (BALB/cSlc-nu/nu; Japan SLC, Inc.), respectively (25). Xenografts were grown for 11–23 weeks and resected, followed by measure for weights (g) and volumes (cm3). Volumes were calculated with the formula as follows: 0.5 × width (cm) × width (cm) × length (cm). Rapamycin (5 mg/kg body weight), PD0325901 (5 mg/kg), and Chrysin (1 mg/kg) were administered into the mice harboring xenografts with intraperitoneal (i.p.) injection. Mice were sacrificed at day 7 to day 9 using isoflurane anesthesia. All of the experimental protocols were approved by the Animal Care and Use Committee at Kyoto Prefectural University of Medicine (M30-57454).

Statistical analysis

The data were presented as means ± SD of multiple samples. Statistical analyses were performed using JMP built-in-modules (SAS Institute Inc.) and Microsoft Excel (Microsoft Japan Co., Ltd.; ref. 26). P < 0.05 and false discovery rate (FDR) q < 0.01 were considered statistically significant.

Generation of methotrexate-resistant PCNSL-derived cells

To know how differences of metabolomic profiles of methotrexate-resistant PCNSL cells compared with nonresistant cells, we first generated two independent methotrexate-resistant PCNSL-derived cells, including methotrexate-resistant TK and HKBML cells (Fig. 1A). For cell viability calculations, the values of inhibitory concentration 50 (IC50) in control TK cells and methotrexate-resistant TK (TK-MTX) cells were 3.73 × 10−8 mol/L and 8.76 × 10−6 mol/L, respectively, showing 234-fold resistance against methotrexate in TK cells. While in HKBML-derived cells, the values of IC50 in the control HKBML cells and methotrexate-resistant HKBML (HKBML-MTX) cells were 1.17 × 10−10 mol/L and 2.14 × 10−8 mol/L, respectively, showing 183-fold resistance against methotrexate in HKBML cells. These results indicate that degrees of methotrexate resistances in both PCNSL cells are similar, whereas the basal resistance of TK is 319-fold compared with that of HKBML.

Figure 1.

Differentially expressed metabolites in methotrexate-resistant primary CNS lymphoma (PCNSL)-derived cell lines TK- and HKBML. A, Generation of the methotrexate-resistant PCNSL-derived TK and HKBML cells. WST-8 cell proliferation assays on the TK- and HKBML-derived cells within an RPMI1640 medium supplemented with 10% fetal calf serum (FCS), and Ham F-12 medium supplemented with 15% FCS, respectively, with diluted methotrexate from 1 × 10−3 to 1 × 10−9 mol m−3 (M). Cell viability is shown. Left, Control and methotrexate-resistant TK. Center, Control and methotrexate-resistant HKBML. Right, IC50 of each cell was calculated. B, Hierarchical clustering analysis (HCA)-heatmap plots of differentially expressed metabolites between methotrexate-resistant TK and HKBML cells. A1–A3, TK (Control); B1–B3, methotrexate-resistant TK; C1–C3, HKBML (Control); D1–D3, methotrexate-resistant HKBML. Color configurations indicate that these metabolites were downregulated (green) and upregulated (red), respectively, compared with the mean of sample. C, Principal component analysis (PCA) plots for metabolites detected between methotrexate-resistant TK and HKBML cells. A1–A3, TK (Control); B1–B3, methotrexate-resistant TK; C1–C3, HKBML (Control); D1–D3, methotrexate-resistant HKBML; PC, principal component. D–I, Differentially expressed metabolites in methotrexate-resistant TK and HKBML cells. The numbers of metabolites detected are represented in Venn diagram. Numbers in the parentheses indicate the number categorized. D, Central carbon metabolism. E, Urea cycle. F, Lipid metabolism. G, Branched-chain amino acids and aromatic amino acids. H, Nucleic acids. I, Coenzyme metabolism. The triplicate samples in each cell were analyzed in the metabolome analysis.

Figure 1.

Differentially expressed metabolites in methotrexate-resistant primary CNS lymphoma (PCNSL)-derived cell lines TK- and HKBML. A, Generation of the methotrexate-resistant PCNSL-derived TK and HKBML cells. WST-8 cell proliferation assays on the TK- and HKBML-derived cells within an RPMI1640 medium supplemented with 10% fetal calf serum (FCS), and Ham F-12 medium supplemented with 15% FCS, respectively, with diluted methotrexate from 1 × 10−3 to 1 × 10−9 mol m−3 (M). Cell viability is shown. Left, Control and methotrexate-resistant TK. Center, Control and methotrexate-resistant HKBML. Right, IC50 of each cell was calculated. B, Hierarchical clustering analysis (HCA)-heatmap plots of differentially expressed metabolites between methotrexate-resistant TK and HKBML cells. A1–A3, TK (Control); B1–B3, methotrexate-resistant TK; C1–C3, HKBML (Control); D1–D3, methotrexate-resistant HKBML. Color configurations indicate that these metabolites were downregulated (green) and upregulated (red), respectively, compared with the mean of sample. C, Principal component analysis (PCA) plots for metabolites detected between methotrexate-resistant TK and HKBML cells. A1–A3, TK (Control); B1–B3, methotrexate-resistant TK; C1–C3, HKBML (Control); D1–D3, methotrexate-resistant HKBML; PC, principal component. D–I, Differentially expressed metabolites in methotrexate-resistant TK and HKBML cells. The numbers of metabolites detected are represented in Venn diagram. Numbers in the parentheses indicate the number categorized. D, Central carbon metabolism. E, Urea cycle. F, Lipid metabolism. G, Branched-chain amino acids and aromatic amino acids. H, Nucleic acids. I, Coenzyme metabolism. The triplicate samples in each cell were analyzed in the metabolome analysis.

Close modal

Metabolome analysis in the methotrexate-resistant PCNSL-derived cells

We considered the difference between methotrexate-resistant and nonresistant PCNSL cells and then conducted metabolome analyses among them including methotrexate-resistant and nonresistant TK and HKBML cells (n = 3; Supplementary Figs. S1 and S2). Hierarchical clustering analysis showed the two distinct subgroups between the methotrexate-resistant and nonresistant cells in TK and HKBML-derived cells (Fig. 1B). PCA also showed compact component sets of metabolites detected in TK-derived cells (86.3% of 188 peaks including 102 cations and 86 anions) and HKBML-derived cells (65.2% of 169 peaks including 98 cations and 71 anions; Fig. 1C). As summarized in the Venn diagrams (Fig. 1DI), most metabolites detected in TK-MTX cells decreased in the urea cycle (Fig. 1E), lipid metabolism (Fig. 1F), branched chain and aromatic amino acids (Fig. 1G), nucleic acids metabolism (Fig. 1H), and coenzyme metabolism (Fig. 1I). Some metabolites detected in HKBML-MTX cells increased in central carbon metabolism (Fig. 1D), urea cycle (Fig. 1E), lipid metabolism (Fig. 1F), and coenzyme metabolism (Fig. 1I), in addition to decreases of most of the metabolites detected in central carbon metabolism (Fig. 1D) and branched chain and aromatic amino acids (Fig. 1G). These findings clearly indicated that most of the metabolites detected in TK-MTX cells decreased, whereas some metabolites detected in HKBML-MTX cells increased compared with the corresponding control cells, suggesting differences in cell type–specific metabolic pathways between the two PCNSL-derived cells.

Cell type–specific metabolic pathways in the methotrexate-resistant PCNSL-derived cells

On the basis of data of discrete metabolites detected, we found that suppression in TK-MTX cells (0.167- to 0.241-fold) and activation in HKBML-MTX cells (1.014-3.273-fold) altered the following pathways: protein metabolism, lipid metabolism, cell function, apoptosis, osmolytes, sugar metabolism, transmethylation, liver disease, and renal disease, compared with the corresponding control cells (Supplementary Table S1). Meanwhile, both TK-MTX (0.248-0.489-fold) and HKBML-MTX cells (0.925-0.619-fold) suppressed electron carrier, purine bases, collagen, arthritis, imidazole compounds, nucleotide sugars, essential amino acid, and methylglyoxal, compared with the corresponding control cells (Supplementary Table S1). Furthermore, based on the data from Human Metabolome Database (HMDB) category, as a representative data, TK-MTX cells were specifically suppressed in arthritis (0.314-fold, P = 0.000411), guanidino compounds (0.186-fold, P = 0.005687), electron carrier (0.329-fold, P = 0.007003), but not in HKBML-MTX cells, compared to the corresponding control cells (Supplementary Table S2). HKBML-MTX cells were specifically activated in renal disease (1.564-fold, P = 0.047601), lipid metabolism (2.07-fold, P = 0.003866), pyrimidine bases (1.158-fold, P = 0.020303), and neuropsychiatric disorder (2.449-fold, P = 0.022399), but these were suppressed in TK-MTX cells, compared with the corresponding control cells (Supplementary Table S2). Summarized cell type–specific metabolites detected in TK-MTX and HKBML-MTX cells, glucose 1-phosphate and fructose 1,6-diphosphate, lactic acid, and acetyl coenzyme A (divalent) decreased specifically in TK-MTX cells (<0.11-fold, P < 0.049), compared with the control TK cells (Supplementary Table S3), indicating that TK-MTX cells are dysregulated in glycolysis. While in HKBML-MTX cells, amino acids including Ala, Trp, His, Phe, and Tyr decreased (<0.5-fold, P < 0.035), but Gln and Asp increased (>1.6-fold, P < 0.047), compared with the control HKBML cells (Supplementary Table S3), indicating that HKBML-MTX cells are dysregulated in amino acid metabolisms.

Glycolysis, ATP synthesis, glutamate metabolism, and mitochondria activity in the methotrexate-resistant PCNSL-derived cells

Considering the abovementioned metabolic profiles, we decided to further investigate glycolysis, ATP synthesis, glutamate metabolism, and mitochondria activity. We reevaluated the amounts of metabolites in the glycolysis pathway using biochemical analysis techniques (Fig. 2AC). The amounts of glucose and ATP significantly increased in both TK-MTX (1.276-fold and 1.613-fold, respectively) and HKBML-MTX (1.102-fold and 1.421-fold, respectively) cells, compared with the corresponding control cells (Fig. 2A). However, glycogen, pyruvate and lactate significantly decreased in both TK-MTX and HKBML-MTX cells, compared with the corresponding control cells (Fig. 2A). LDH activities were upregulated in both TK-MTX (1.465-fold) and HKBML-MTX (1.83-fold) cells with statistical significances, compared with the corresponding control cells (Fig. 2B and C). These results suggest that cellular glycogen storages were released, and concentrated glucose was processed into pyruvate and also to lactate by LDH, and then the ATP level increased. The amount of glutamate increased in TK-MTX cells (1.303-fold, P < 0.01), but not in HKBML-MTX. The α-keto-glutamate and the glutamate dehydrogenase (GDH) activities significantly decreased in both TK-MTX (0.585-fold and 0.969-fold, respectively) and HKBML-MTX (0.72-fold and 0.603-fold, respectively) cells, compared with the corresponding control cells (Fig. 2D). Furthermore, the activities of mitochondrial membrane potentials and the amount of Fe(II) oxides at the Fe/S cluster into mitochondrial lumen significantly decreased in both TK-MTX (0.67-fold and 0.95-fold, respectively) and HKBML-MTX (0.53-fold and 0.9-fold, respectively) cells, compared with the corresponding control cells (Fig. 2E). These results suggest that glutamate metabolism that provides glutamate as fuels into tricarboxylic acid (TCA) cycle was downregulated in both TK-MTX and HKBML-MTX with decreased GDH activities and suppressed mitochondrial activities. Moreover, glycolysis is more dominant than aerobic metabolism including TCA cycle in both methotrexate-resistant PCNSL cells.

Figure 2.

Enhanced glycolysis and ATP production but not glutamate metabolism and mitochondria activity in the methotrexate-resistant PCNSL-derived cell lines TK and HKBML. A, Glycolysis, including glucose, glycogen, pyruvate, and lactate, and ATP. Values are normalized by values in 1 × 104 cells in each. B, Cell number-dependent LDH activity in the methotrexate-resistant TK and HKBML cells were measured by 490 nm absorbance. C, Fold activities of LDH in the methotrexate-resistant TK and HKBML cells, compared with the corresponding controls. D, Glutamate metabolism, including glutamate and α-ketogulutamate, and glutamate dehydrogenase (GDH) activity. GDH activity was estimated by the unit (U). E, Mitochondria activity (MitoBright Red) and Fe2+ (Mito-FerroGreen). Values are normalized by values in 1 × 104 cells in both. RFU, relative fluorescence unit. Numbers in parentheses on the top of bars of methotrexate-resistant cells indicate fold differences compared with the values of controls. **, P < 0.01; *, P < 0.05; n.s., not significant. The experiments were repeated at least once with similar results.

Figure 2.

Enhanced glycolysis and ATP production but not glutamate metabolism and mitochondria activity in the methotrexate-resistant PCNSL-derived cell lines TK and HKBML. A, Glycolysis, including glucose, glycogen, pyruvate, and lactate, and ATP. Values are normalized by values in 1 × 104 cells in each. B, Cell number-dependent LDH activity in the methotrexate-resistant TK and HKBML cells were measured by 490 nm absorbance. C, Fold activities of LDH in the methotrexate-resistant TK and HKBML cells, compared with the corresponding controls. D, Glutamate metabolism, including glutamate and α-ketogulutamate, and glutamate dehydrogenase (GDH) activity. GDH activity was estimated by the unit (U). E, Mitochondria activity (MitoBright Red) and Fe2+ (Mito-FerroGreen). Values are normalized by values in 1 × 104 cells in both. RFU, relative fluorescence unit. Numbers in parentheses on the top of bars of methotrexate-resistant cells indicate fold differences compared with the values of controls. **, P < 0.01; *, P < 0.05; n.s., not significant. The experiments were repeated at least once with similar results.

Close modal

Candidate pathways regulating glycolysis in the methotrexate-resistant PCNSL-derived cells

We next examined the upstream signaling regulating glycolysis in the methotrexate-resistant PCNSL-derived cells. First, we investigated cellular signaling pathways in proliferation (Fig. 3, Supplementary Fig. S3, see Supplementary text). Growth suppression of TK cells with 1.6 μmol/L LY294002 as a PI3K inhibitor (0.45-fold, P < 0.05) was recovered by acquired methotrexate resistance, whereas both control HKBML and HKBML-MTX cells were suppressed by LY294002 (0.8-fold and 0.76-fold, P < 0.05, respectively; Fig. 3A and B). Growth suppression of TK cells with 0.4 μmol/L PD0325901 (0.66-fold, P < 0.05), a MAPK inhibitor, was also recovered by acquired methotrexate resistance, whereas HKBML-MTX cells were sensitive to PD0325901 compared with the control HKBML cells (0.73-fold, P < 0.05; Fig. 3C and D). Besides, growth suppression of TK cells with 2.1 μmol/L rapamycin as a mTOR inhibitor (0.74-fold, P < 0.01) was also recovered by acquired methotrexate resistance, whereas HKBML-MTX cells showed a sensitivity for rapamycin, compared with the control HKBML cells (0.78-fold, P < 0.05; Fig. 3E and F). These results suggest that PI3K, MAPK, and mTOR signaling pathways are activated in the TK-MTX cells, but not in HKBML-MTX cells, thus the inhibitors are hard to suppress the growth of TK-MTX cells (1.2-8.57-fold, P < 0.01), but function in the growth suppression of HKBML-MTX cells (0.06-0.84-fold, P < 0.01), compared with the corresponding control cells (Supplementary Fig. S4). However, such malformed growth suppression was not associated with apoptosis and/or cellular senescence because of low signals for caspase-3/7 activity (Supplementary Fig. S5A and S5B) and senescence-associated (SA)-β-galactosidase activity (Supplementary Fig. S5C and S5D), compared with the appropriate positive controls.

Figure 3.

Inhibitors for MEK, PI3K, and mTOR induce suppression of cell growth in the methotrexate-resistant TK and HKBML cells. Dose dependency of drugs, including LY294002 as PI3K inhibitor (A), PD0325901 as a MEK inhibitor (C), and rapamycin as an mTOR inhibitor in the methotrexate-resistant TK and HKBML cells (E). Endo-point suppression of cell growth of the methotrexate-resistant TK and HKBML cells cultured for 4 days with drugs, including LY294002 (B), PD0325901 (D), and rapamycin (F). Cells were cultured for 4 days with diluted drugs. WST-8 cell proliferation assays were measured by 450 nm absorbance and presented by growth ratio compared with values of the appropriate control samples. Numbers in parentheses on the top of bars of methotrexate-resistant cells indicate fold differences compared with the values of controls (**, P < 0.01; *, P < 0.05; n.s., not significant). The experiments were repeated at least once with similar results.

Figure 3.

Inhibitors for MEK, PI3K, and mTOR induce suppression of cell growth in the methotrexate-resistant TK and HKBML cells. Dose dependency of drugs, including LY294002 as PI3K inhibitor (A), PD0325901 as a MEK inhibitor (C), and rapamycin as an mTOR inhibitor in the methotrexate-resistant TK and HKBML cells (E). Endo-point suppression of cell growth of the methotrexate-resistant TK and HKBML cells cultured for 4 days with drugs, including LY294002 (B), PD0325901 (D), and rapamycin (F). Cells were cultured for 4 days with diluted drugs. WST-8 cell proliferation assays were measured by 450 nm absorbance and presented by growth ratio compared with values of the appropriate control samples. Numbers in parentheses on the top of bars of methotrexate-resistant cells indicate fold differences compared with the values of controls (**, P < 0.01; *, P < 0.05; n.s., not significant). The experiments were repeated at least once with similar results.

Close modal

Subsequently, we examined the amounts of glucose, pyruvate, lactate, and ATP, and the LDH activity in the TK-MTX and HKBML-MTX cells with inhibitors (Fig. 4; Supplementary Fig. S6, see Supplementary text). TK-MTX cells treated with PD0325901 showed decreased glucose (0.74-fold, P < 0.01), compared with the control TK cells (Fig. 4A). HKBML-MTX cells showed increased glucose (1.36-fold, P < 0.01), and furthermore, LY29402 and PD0325901 also increased glucose concentration (1.24-fold, P < 0.05 and 1.07-fold, P < 0.01, respectively), whereas rapamycin decreased that in HKBML-MTX cells (0.83-fold, P < 0.01), compared with the control HKBML cells (Fig. 4B). TK-MTX cells showed decreased pyruvate (0.44-fold, P < 0.01), whereas LY294002, PD0325901, and rapamycin increased pyruvate concentration in TK-MTX cells (8.94-fold, P < 0.01; 1.75-fold, P < 0.05; and 16.41-fold, P < 0.01, respectively), compared with the control TK cells (Fig. 4C). HKBML-MTX cells showed decreased pyruvate (0.44-fold, P < 0.01), whereas rapamycin increased pyruvate concentration (4.16-fold, P < 0.01) in HKBML-MTX cells, compared with the control HKBML cells (Fig. 4D). TK-MTX cells showed decreased lactate (0.34-fold, P < 0.01), and furthermore, LY294002, PD0325901, and rapamycin also decreased lactate concentration in TK-MTX cells (0.13-fold, P < 0.01; 0.29-fold, P < 0.01; and 0.03-fold, P < 0.01, respectively) compared with the control TK cells (Fig. 4E). HKBML-MTX cells showed decreased lactate (0.51-fold, P < 0.05), and LY294002 also decreased lactate concentration in HKBML-MTX cells (0.66-fold, P < 0.05), compared with the control HKBML cells (Fig. 4F). LDH activities were also tested with the inhibitors (Fig. 4G and H). The LDH activity in TK-MTX cells was upregulated (2.83-fold, P < 0.01), whereas LY294002, PD0325901, and rapamycin showed the reduced activities of LDH in TK-MTX cells (0.14-fold, P < 0.01; 0.42-fold, P < 0.01; and 0.23-fold, P < 0.01, respectively), compared with the control TK cells (Fig. 4G). Similarly, The LDH activity in HKBML-MTX cells was upregulated (1.66-fold, P < 0.01), whereas LY294002, PD0325901, and rapamycin showed the reduced activities of LDH in HKBML-MTX cells (0.52-fold, P < 0.01; 0.45-fold, P < 0.05; and 0.67-fold, P < 0.01, respectively), compared with the control HKBML cells (Fig. 4H). We also examined ATP synthesis with the inhibitors (Fig. 4I and J). The ATP increased in TK-MTX cells (3.22-fold, P < 0.01), and LY294002 also increased ATP concentration in TK-MTX cells (2.57-fold, P < 0.01), compared with the control TK cells (Fig. 4I). HKBML-MTX cells showed increased ATP (1.53-fold, P < 0.01), and LY294002 also increased ATP concentration in HKBML-MTX cells (1.65-fold, P < 0.05), whereas PD0325901 and rapamycin decreased ATP concentration (0.48-fold, P < 0.01, and 0.01-fold, P < 0.01, respectively), compared with the control HKBML cells (Fig. 4J). Considering the abovementioned results, these data indicate that both TK-MTX and HKBML-MTX cells tend to upregulate glycolysis with LDH activity and ATP synthesis, whereas the inhibition of PI3K, MAPK, and mTOR signaling pathways especially compete the effects of methotrexate resistance in such metabolic pathways in the TK-MTX cells, suggesting that TK cells have a potential methotrexate resistance than HKBML cells, which is consistent with the results in cell proliferation, as already shown in Fig. 1A.

Figure 4.

Cell-specific manners for glycolysis regulation, including glucose, pyruvate, and lactate, and for ATP synthesis in the methotrexate-resistant PCNSL-derived TK and HKBML cells cocultured with signaling inhibitors for PI3K, MEK, and mTOR. in the methotrexate-resistant PCNSL-derived cell lines TK and HKBML. A and B, Glucose. C and D, Pyruvate. E and F, Lactate. G and H, LDH activity. I and J, ATP. Values are normalized by values in 1 × 106 TK cells (A, C, E, G, and I) and 5 × 105 HKBML cells (B, D, F, H, and J), respectively. LDH activity is expressed by the normalized absorbance at 450 nm (1 × 106 TK cells and 5 × 105 HKBML cells). Numbers in parentheses on the top of bars of methotrexate-resistant cells indicate fold differences compared with the values of appropriate controls (**, P < 0.01; *, P < 0.05; n.s., not significant). The experiments were repeated at least once with similar results.

Figure 4.

Cell-specific manners for glycolysis regulation, including glucose, pyruvate, and lactate, and for ATP synthesis in the methotrexate-resistant PCNSL-derived TK and HKBML cells cocultured with signaling inhibitors for PI3K, MEK, and mTOR. in the methotrexate-resistant PCNSL-derived cell lines TK and HKBML. A and B, Glucose. C and D, Pyruvate. E and F, Lactate. G and H, LDH activity. I and J, ATP. Values are normalized by values in 1 × 106 TK cells (A, C, E, G, and I) and 5 × 105 HKBML cells (B, D, F, H, and J), respectively. LDH activity is expressed by the normalized absorbance at 450 nm (1 × 106 TK cells and 5 × 105 HKBML cells). Numbers in parentheses on the top of bars of methotrexate-resistant cells indicate fold differences compared with the values of appropriate controls (**, P < 0.01; *, P < 0.05; n.s., not significant). The experiments were repeated at least once with similar results.

Close modal

Differential gene expression for PI3K, MAPK, and mTOR signaling pathways in the methotrexate-resistant PCNSL-derived cells

We furthermore investigated changes in gene expression related to PI3K signaling (PIK3R1, PIK3CA, PDK1, PTEN, AKT1, and GSK3B), cell proliferation (MYC, PCNA, and MKI67), mTOR complex formation (MTOR, RPTOR, RICTOR, and DEPTOR), glucose metabolism and transport (G6PT1, G6PC, and GLUT2), and transcription factors related to such signaling pathways (BCL2/6, FOXO1, GATA1/2/3, HIF1A, NFKB1/2, and TP53) in the methotrexate-resistant PCNSL-derived cells (Fig. 5A and B; Supplementary Table S4). The expression levels of the following genes were significantly upregulated in the TK-MTX cells compared with the control TK cells: MYC (3.96-fold), PIK3R1 (3.76-fold), MTOR (2.69-fold), BCL6 (2.5-fold), MKI67 (2.02-fold), RPTOR (1.81-fold), GATA2 (1.77-fold), NFKB2 (1.71-fold), PIK3CA (1.66-fold), TP53 (1.49-fold), PTEN (1.47-fold), and G6PT1 (1.42-fold). However, the expression levels of the following genes were significantly downregulated in the TK-MTX cells as compared with the control TK cells: GLUT2 (0.32-fold), NFKB1 (0.42-fold), PCNA (0.65-fold), and FOXO1 (0.66-fold; Fig. 5A). As compared with the control HKBML cells, the following proteins were significantly upregulated in the HKBML-MTX cells: MTOR (5.41-fold), BCL6 (3.08-fold), HIF1A (2.29-fold), FOXO1 (2.13-fold), GLUT2 (1.99-fold), MYC (1.62-fold), G6PC (1.47-fold), BCL2 (1.36-fold), MKI67 (1.13-fold), G6PT1 (1.13-fold), and NFKB2 (1.02-fold). However, the expression levels of the following genes were significantly downregulated in the HKBML-MTX cells compared with in the control HKBML cells: PIK3R1 (0.14-fold), DEPTOR (0.65-fold), AKT1 (0.69-fold), RPTOR (0.76-fold), and GATA3 (0.78-fold) (Fig. 5B). Summarizing these expression data, the upregulation of MAPK signaling, cell proliferation (MYC, BCL6, and MKI67), and B-cell differentiation (NFKB2) was considered to contribute to the growth of TK and HKBML cells reprogrammed by methotrexate (Fig. 5C and D). The upregulation of PI3K signaling (PIK3R1 and PIK3CA), mTOR signaling (mTOR and RPTOR), and multipotent transcription factor GATA2, and the downregulation of FOXO1 for proapoptotic signaling likely contributed to the growth of TK-MTX cells (Fig. 5C). The upregulation of glucose transport (GLUT2, G6PC, and G6PT1) and hypoxia transcription factor HIF1A contributed specifically to the growth of HKBML-MTX cells (Fig. 5D).

Figure 5.

Cell type–specific signaling to glycolysis and cell growth in the methotrexate-resistant PCNSL cells. A and B, Differential expression of the genes related to PTEN/PI3K/AKT/mTOR signaling, B-cell development, and cell proliferation in the methotrexate-resistant PCNSL cells. A, TK cells. B, HKBML cells. Fold changes of gene expression in the methotrexate-resistant cells compared with the corresponding control cells are presented. *, P < 0.05. C and D, Schematics of cell type–specific manners for glycolysis and cell growth in the methotrexate-resistant TK and HKBML cells. C, The methotrexate-resistant TK cells activated by PI3K/RAS/MAPK and PI3K/AKT/mTOR signaling pathways are allowed cell growth under a treatment of drugs, including LY294002 (PI3K inhibitor), PD0325901 (MEK inhibitor), and rapamycin (mTOR inhibitor). The PI3K, MAPK, and mTOR signaling is enhanced in the methotrexate-resistant TK cells. D, The methotrexate-resistant HKBML cells obtain to susceptibility for PD0325901 and rapamycin. The MYC pathway downstream at MAPK and mTOR multiple growth signaling pathways is suppressed, but other signaling pathway activates the MYC expression and glycolysis in the methotrexate-resistant HKBML cells. The experiments were repeated at least once with similar results.

Figure 5.

Cell type–specific signaling to glycolysis and cell growth in the methotrexate-resistant PCNSL cells. A and B, Differential expression of the genes related to PTEN/PI3K/AKT/mTOR signaling, B-cell development, and cell proliferation in the methotrexate-resistant PCNSL cells. A, TK cells. B, HKBML cells. Fold changes of gene expression in the methotrexate-resistant cells compared with the corresponding control cells are presented. *, P < 0.05. C and D, Schematics of cell type–specific manners for glycolysis and cell growth in the methotrexate-resistant TK and HKBML cells. C, The methotrexate-resistant TK cells activated by PI3K/RAS/MAPK and PI3K/AKT/mTOR signaling pathways are allowed cell growth under a treatment of drugs, including LY294002 (PI3K inhibitor), PD0325901 (MEK inhibitor), and rapamycin (mTOR inhibitor). The PI3K, MAPK, and mTOR signaling is enhanced in the methotrexate-resistant TK cells. D, The methotrexate-resistant HKBML cells obtain to susceptibility for PD0325901 and rapamycin. The MYC pathway downstream at MAPK and mTOR multiple growth signaling pathways is suppressed, but other signaling pathway activates the MYC expression and glycolysis in the methotrexate-resistant HKBML cells. The experiments were repeated at least once with similar results.

Close modal

Phosphorylation of the PI3K–AKT–mTOR and MAPK were confirmed with protein analyses, suggesting that these signaling pathways are activated in the methotrexate-resistant PCNSL cells (Supplementary Figs. S7 and S8, see Supplementary text). Furthermore, global expression analyses demonstrated that downregulated function of membrane activity, glucose, and central carbon metabolism and upregulated stem cell property exist in HKBML-MTX, and downregulated function of negative regulation of MAPK, ECM disassembly, B-cell receptor signaling, and NK-cell–mediated cytotoxicity and upregulated function of glycosylation of glycoproteins and EGF-like signaling pathway are served in TK-MTX (Supplementary Fig. S9; Supplementary Table S5, see Supplementary text). Besides, the GO analyses denoted that tissue stem cell function and immune system process are activated in both HKMBL-MTX and TK-MTX and showed that downregulated signaling of endogenous galectin-1 exists in HKBML-MTX. In the GSEA, TK-MTX also showed the downregulation of cell projection, neural development, and blood cell differentiation (Supplementary Table S6, see Supplementary text).

Tumorigenicity of PCNSL xenografts with blockades for mTOR and HIF-1α pathways in vivo

To investigate tumorigenicity of PCNSL cells in vivo, we transplanted PCNSL-derived cells into subcutaneous tissues in the back of nude mice. Survival rates of the xenografts of the cells were 11.62% (n = 5/43) in TK and 25% (n = 13/52) in HKBML, suggestive of low viabilities and effects of tumor microenvironments. The xenografts of PCNSL-derived cells were grown in animals and examined drug sensitivities using rapamycin (mTOR inhibitor), PD0325901 (MEK inhibitor), and Chrysin (HIF1A inhibitor; Fig. 6; Supplementary Fig. S10). The measurements in tumor weight and volume clarified that TK-MTX was engrafted than TK (1.47-fold; **, P = 0.0078 in weight, 1.62-fold; *, P = 0.0167 in volume), but the both TK and TK-MTX were suppressed with rapamycin (Fig. 6AC). While, HKBML could be suppressed by rapamycin (0.08-fold; P = 0.1381 in weight, 0.31-fold; *, P = 0.0366 in volume), but HKBML-MTX sustained prominent tumors (Fig. 6D). In addition, HKMBL-MTX evaded tumor suppression by rapamycin (8.42-fold; ***, P = 5.17 × 10−5 in weight, 8.5-fold; *, P = 0.0201 in volume) (Fig. 6E and F). Besides, HKBML-MTX showed a susceptibility to PD0325901 in tumor weight (0.2-fold; **, P = 0.0028). Furthermore, HKBML showed a Chrysin resistance in tumor volume (5.61-fold; **, P = 0.0039), whereas HKBML-MTX acquired a Chrysin sensitivity (0.09-fold; P = 0.0635 in weight, 0.43-fold; *, P = 0.0204 in volume; Fig. 6DF). These results indicated that the mTOR pathway inhibition was effective for suppression of TK, TK-MTX, and HKBML in vivo, but not in HKBML-MTX, and that the in vivo growth suppression of HKBML-MTX could be caused by the HIF-1α pathway inhibition.

Figure 6.

The methotrexate-resistant PCNSL-derived cells acquire drug sensitivities and resistances in the tumor xenograft nude mice model. A–C, The xenografts of TK and TK-MTX were grown in nude mice (control: n = 3, case: n = 2). A, The representative images of the xenografts of TK and TK-MTX with or without rapamycin (mTOR inhibitor). B, Tumor weight (g) of the TK-derived xenografts. C, Tumor volume (cm3) of the TK-derived xenografts. D–F, The xenografts of HKBML and HKBML-MTX were grown in nude mice (control: n = 3, case: n = 3–4). D, The representative images of the xenografts of HKBML and HKBML-MTX with rapamycin (n = 3), PD0325901 (MEK inhibitor; n = 4) and Chrysin (HIF1A inhibitor; n = 3). E, Tumor weight (g) of the HKBML-derived xenografts. F, Tumor volume (cm3) of the HKBML-derived xenografts. Mice were sacrificed at day 7 to day 9 after drug administration. Scale bars, 5 mm (*, P < 0.05; **, P < 0.01; ***, P < 0.001).

Figure 6.

The methotrexate-resistant PCNSL-derived cells acquire drug sensitivities and resistances in the tumor xenograft nude mice model. A–C, The xenografts of TK and TK-MTX were grown in nude mice (control: n = 3, case: n = 2). A, The representative images of the xenografts of TK and TK-MTX with or without rapamycin (mTOR inhibitor). B, Tumor weight (g) of the TK-derived xenografts. C, Tumor volume (cm3) of the TK-derived xenografts. D–F, The xenografts of HKBML and HKBML-MTX were grown in nude mice (control: n = 3, case: n = 3–4). D, The representative images of the xenografts of HKBML and HKBML-MTX with rapamycin (n = 3), PD0325901 (MEK inhibitor; n = 4) and Chrysin (HIF1A inhibitor; n = 3). E, Tumor weight (g) of the HKBML-derived xenografts. F, Tumor volume (cm3) of the HKBML-derived xenografts. Mice were sacrificed at day 7 to day 9 after drug administration. Scale bars, 5 mm (*, P < 0.05; **, P < 0.01; ***, P < 0.001).

Close modal

Comparing the exo-metabolomics of differentiating murine B lymphoma cells and primary B cells with one-dimensional proton nuclear magnetic resonance (NMR) spectroscopy and LC/MS, lactate production increased with reduced essential amino acids. Moreover, immunoglobulin secretion was paralleled by Ala, Glu, and Gln as carbon and energy sources (27). With antigen-binding small B lymphocytes, a proteomic metamorphosis is activated to generate antibody-secreting cells (28). Methotrexate is an antifolate that is a DHFR inhibitor in purine and thymidine syntheses (5). Meanwhile, a serine hydroxymethyltransferase (SHMT) converts Ser into Gly and a tetrahydrofolate-bound one-carbon unit for purine and thymidine synthesis and for cell growth (29). In Burkitt lymphoma and DLBCL, gas chromatography-mass spectrometry (GC-MS) and nanoLC-sequential window acquisition of all theoretical (SWATH)-MS clarify decreased pyruvic acid content, and downregulated glycolysis but upregulated one-carbon metabolism in Burkitt lymphoma, compared with DLBCL (30). In anaplastic large cell lymphoma, integrated phosphoproteomic and metabolomic analyses reveal that constitutively active tyrosine kinase nucleophosmin-anaplastic lymphoma kinase (NPM-ALK) induces a metabolic shift toward aerobic glycolysis, lactate production, and biomass production. The metabolic shift is mediated by ALK phosphorylation at Y105 of the tumor-specific isoform of pyruvate kinase 2 (PYK2) protein, resulting in decreased enzymatic activity (31).

Altered metabolites and metabolomic pathways are undoubtedly biomarkers for cancer development and progression (32). In serum lipid analysis on a subset of individuals from a population-based NHL case–control study from 100 patients with DLBCL, significantly elevated levels of endocannabinoid and 2-arachidonoylglycerol (2-AG) were detected in the serum of DLBCL patients (33). NMR-based serum metabolomics identifies the following metabolites as high-risk signatures for patients with DLBCL: Lys and Arg; the degradation product cadaverine and a compound in oxidative stress (2-hydroxybutyrate) in the refractory/early relapse group (n = 27), and Asp, Val and ornithine, and a metabolite in the glutathione cycle, pyroglutamate in long-term progression-free patients (n = 60) (34). In Burkitt lymphoma mice, NMR-based metabolomics demonstrate that the serum metabolites are altered during the metabolism of energy, amino acids, fatty acids, and choline phospholipids (35). The diagnostic potential of differential metabolites also clarifies potential biomarkers including Glu, glycerol, and choline with a high diagnostic accuracy; in contrast, Ile, Leu, pyruvate, Lys, α-ketoglutarate, betaine, Gly, creatine, Ser, lactate, Tyr, Phe, His and formate, discriminating Burkitt lymphoma mice from normal mice (35). In cutaneous T-cell lymphoma (CTCL), a class of NHL, and a heterogeneous group of skin-homing T-cell neoplasms, ultrahigh performance liquid chromatography–quadrupole (UHPLC-Q) TOF/MS and further LC/MS-MS determined the 36 potential biomarkers associated with CTCL, which are derived from CTCL plasma metabolic perturbations (36). In skin and plasma of CTCL mice, UHPLC-QTOF/MS also demonstrate that increased l-glutamate and decreased adenosine monophosphate are the most essential metabolic pathways of CTCL tumors and tumor adjacent skins (37). Data from GC-MS and UHPLC-QTOF/MS show that glycerophospholipid metabolism, tryptophan metabolism, and purine metabolism are altered pathways in serum samples from 31 patients with CTCL (38).

Metabolomics is a comprehensive method to identify and quantify metabolites from biological samples including cells (19), sera (35), and urines (18). However, few studies discuss glycolysis in CNS lymphoma. Here we generated two independent methotrexate-resistant PCNSL-derived cells as models of refractory and relapse-acquired methotrexate resistance, designated as TK-MTX and HKBML-MTX, and then conducted metabolomic analyses to know how changes in endogenous metabolites in methotrexate-resistant PCNSL-derived cells affect their viability. The metabolomic data from a capillary electrophoresis (CE)-connected electrospray ionization-time-of-flight mass spectrometry (ESI-TOFMS) and CE-MS/MS, and the annotation by HMDB demonstrated cell type–specific metabolic pathways with changes of glycolysis and amino acid metabolism in TK-MTX and HKBML-MTX cells, respectively, although noncharacterized metabolites should be addressed. Furthermore, we verified the differentially concentrated metabolites using biochemical methods in vitro, which were focused on glycolysis (with LDH), ATP synthesis, glutamate metabolism (with GDH), and mitochondrial activity, based on the metabolomic data. Consequently, TK cells were more dominant than HKBML cells in proliferation following methotrexate exposure, whereas the reprogrammed TK-MTX cells were governed by PI3K/AKT/mTOR and RAS/MAP-Kinase signaling pathways. Conversely, the HKBML-MTX cells acquired sensitivities at mTOR and MAPK signaling pathways without proapoptotic and cellular senescence activities. Besides, the genes related to the signaling pathways including PI3K/AKT/mTOR, RAS/MAP-Kinase and the transcription factors involved were also dysregulated in the methotrexate-resistant PCNSL-derived cells (e.g., increases of MYC, PIK3R1, PIK3CA, MTOR, and RPTOR in TK-MTX cells and increases of MYC, G6PT1, G6PC, GLUT2, HIF1A, and BCL6 in HKBML-MTX cells). Considering suppressive electron carriers in both methotrexate-resistant PCNSL-derived cells (Supplementary Table S1), these observations suggest that PCNSL cells are reprogramed by methotrexate into the cells harboring the metabolism via aerobic glycolysis than oxidative phosphorylation pathway, such as the Warburg effect (39). TK cells were sensitive to oxidative stress but HKBML cells were resistant (Supplementary Figs. S11–S13, see Supplementary text). Furthermore, HKBML had a potential activity of reactive oxygen species (ROS) and enhanced HIF1α activity under oxidative stress (Supplementary Fig. S11). Thus, there would be cell type–specific and oxidative stress–dependent signaling pathways via HIF1α and c-MYC for glycolysis in differential manners (Supplementary Fig. S12). On the other hand, of great interest was that HKBML-MTX cells showed sensitivity for mTOR inhibitors and MEK inhibitors. This suggests a possibility that, after diagnosing HKBML cell types with metabolomic biomarkers, mTOR and/or MAPK signaling pathways could be promising therapeutic targets for growth suppression of CNS lymphomas with refractory and relapse-acquired methotrexate resistances. While, a recent study clarified that TK-MTX was sensitive to bortezomib than parent TK, but HKBML was not affected (20). The qPCR results from PCNSL tissues clarified that the subgroups with higher expression of MTOR and PIK3R1 showed poor prognoses, also suggesting that PI3K and mTOR pathways are crucial for biology in the methotrexate-resistant PCNSL with poor survival (Supplementary Table S7; Supplementary Figs. S14–S16, see Supplementary text).

We also tested in vivo tumorigenicity of the PCNSL-derived cells and their drug resistance/sensitivity using nude mice. The results demonstrated that the mTOR inhibition caused the growth suppression of the xenografts of the PCNSL-derived cells except for HKBML-MTX, although TK-MTX and HKBML-MTX were resistant and sensitive to rapamycin, respectively, in vitro. Moreover, PD0325901 MEK inhibitor slightly failed growth suppression of HKBML and HKBML-MTX in vivo. These results are not consistent with the results from in vitro experiments in the current study. However, such discrepancies might be explained by the differences of tumor microenvironments in vitro and in vivo, which should await future studies. The microenvironments might also be associated with a low survival rate of xenografts in nude mice. On the other hand, the HIF-1α inhibition by Chrysin clearly induced growth suppression of the HKBML-MTX xenografts, but not in the HKBML. This is consistent with the results from in vitro experiments, such as the antioxidative stress cell growth, ROS activity, HIF1A expression, high GSH/GSSG ratio, and glutathione reductase activity. These in vivo results also suggested cell type–specific manners for cell survival and growth and signaling pathways in PCNSL. Thus, we must diagnose such cell-types before treatments in future. This study would be a hint for understanding the signaling at the methotrexate and folate metabolisms, drug discovery, and development for de novo biomarkers and pathways in PCNSL.

No potential conflicts of interest were disclosed.

Conception and design: Y. Takashima, R. Yamanaka

Development of methodology: R. Yamanaka

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Y. Takashima, A. Hayano, R. Yamanaka

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y. Takashima, R. Yamanaka

Writing, review, and/or revision of the manuscript: Y. Takashima, R. Yamanaka

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Y. Takashima, A. Hayano, R. Yamanaka

Study supervision: R. Yamanaka

This work was supported by JSPS KAKENHI (grant numbers 16H05441, to R. Yamanaka; 18K09001, to Y. Takashima). The authors thank Dr. Junya Fukai (Wakayama Medical University), Dr. Yasuo Iwadate (Chiba University), Dr. Koji Kajiwara (Yamaguchi University), and Dr. Hiroaki Hondoh (Toyama Prefectural Central Hospital) for providing PCNSL samples and clinical data.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
WHO classification of tumours of the central nervous system
. In:
Louis
DN
,
Ohgaki
H
,
Wiestler
OD
,
Cavenee
WK
, editors.
World Health Organization; International Agency for Research on Cancer. Revised 4th edition
.
Lyon, France
:
International Agency for Research on Cancer
; 
2016
.
2.
Louis
DN
,
Perry
A
,
Reifenberger
G
,
von Deimling
A
,
Figarella-Branger
D
,
Cavenee
WK
, et al
The 2016 World Health Organization classification of tumors of the central nervous system: a summary
.
Acta Neuropathol
. 
2016
;
131
:
803
20
.
3.
Ricard
D
,
Idbaih
A
,
Ducray
F
,
Lahutte
M
,
Hoang-Xuan
K
,
Delattre
JY
. 
Primary brain tumours in adults
.
Lancet
. 
2012
;
379
:
1984
96
.
4.
Chabner
BA
,
Allegra
CJ
,
Curt
GA
,
Clendeninn
NJ
,
Baram
J
,
Koizumi
S
, et al
Polyglutamation of methotrexate. Is methotrexate a prodrug?
J Clin Invest
. 
1985
;
76
:
907
12
.
5.
Rushworth
D
,
Mathews
A
,
Alpert
A
,
Cooper
LJ
. 
Dihydrofolate reductase and thymidylate synthase transgenes resistant to methotrexate interact to permit novel transgene regulation
.
J Biol Chem
. 
2015
;
290
:
22970
6
.
6.
Goecke
IA
,
Alvarez
C
,
Henríquez
J
,
Salas
K
,
Molina
ML
,
Ferreira
A
, et al
Methotrexate regulates the expression of glucocorticoid receptor alpha and beta isoforms in normal human peripheral mononuclear cells and human lymphocyte cell lines in vitro
.
Mol Immunol
. 
2007
;
44
:
2115
23
.
7.
Kasenda
B
,
Ferreri
AJ
,
Marturano
E
,
Forst
D
,
Bromberg
J
,
Ghesquieres
H
, et al
First-line treatment and outcome of elderly patients with primary central nervous system lymphoma (PCNSL)–a systematic review and individual patient data meta-analysis
.
Ann Oncol
. 
2015
;
26
:
1305
13
.
8.
Kawaguchi
A
,
Iwadate
Y
,
Komohara
Y
,
Sano
M
,
Kajiwara
K
,
Yajima
N
, et al
Gene expression signature-based prognostic risk score in patients with primary central nervous system lymphoma
.
Clin Cancer Res
. 
2012
;
18
:
5672
81
.
9.
Iwadate
Y
,
Suganami
A
,
Ikegami
S
,
Shinozaki
N
,
Matsutani
T
,
Tamura
Y
, et al
Non-deep-seated primary CNS lymphoma: therapeutic responses and a molecular signature
.
J Neurooncol
2014
;
117
:
261
8
.
10.
Yamanaka
R
. 
Salvage therapy for primary central nervous system lymphoma
. In:
Yamanaka
R
, editor.
Primary central nervous system lymphoma (PCNSL): incidence, management and outcomes
.
New York, NY
:
Nova Science Publishers
;
2016
. p.
175
87
.
11.
Reni
M
,
Ferreri
AJ
,
Villa
E
. 
Second-line treatment for primary central nervous system lymphoma
.
Br J Cancer
. 
1999
;
79
:
530
4
.
12.
Koppenol
WH
,
Bounds
PL
,
Dang
CV
. 
Otto Warburg's contributions to current concepts of cancer metabolism
.
Nat Rev Cancer
. 
2011
;
11
:
325
37
.
13.
Hu
J
,
Locasale
JW
,
Bielas
JH
,
O'Sullivan
J
,
Sheahan
K
,
Cantley
LC
, et al
Heterogeneity of tumor-induced gene expression changes in the human metabolic network
.
Nat Biotechnol
. 
2013
;
31
:
522
9
.
14.
Cairns
RA
,
Harris
IS
,
Mak
TW
. 
Regulation of cancer cell metabolism
.
Nat Rev Cancer
. 
2011
;
11
:
85
95
.
15.
Ward
PS
,
Thompson
CB
. 
Metabolic reprogramming: a cancer hallmark even Warburg did not anticipate
.
Cancer Cell
. 
2012
;
21
:
297
308
.
16.
Yun
J
,
Rago
C
,
Cheong
I
,
Pagliarini
R
,
Angenendt
P
,
Rajagopalan
H
, et al
Glucose deprivation contributes to the development of KRAS pathway mutations in tumor cells
.
Science
. 
2009
;
325
:
1555
9
.
17.
Metallo
CM
,
Gameiro
PA
,
Bell
EL
,
Mattaini
KR
,
Yang
J
,
Hiller
K
, et al
Reductive glutamine metabolism by IDH1 mediates lipogenesis under hypoxia
.
Nature
. 
2011
;
481
:
380
4
.
18.
Yoo
BC
,
Kong
SY
,
Jang
SG
,
Kim
KH
,
Ahn
SA
,
Park
WS
, et al
Identification of hypoxanthine as a urine marker for non-Hodgkin lymphoma by low-mass-ion profiling
.
BMC Cancer
. 
2010
;
10
:
55
.
19.
Pera
B
,
Krumsiek
J
,
Assouline
SE
,
Marullo
R
,
Patel
J
,
Phillip
JM
, et al
Metabolomic profiling reveals cellular reprogramming of B-cell lymphoma by a lysine deacetylase inhibitor through the choline pathway
.
EBioMedicine
. 
2018
;
28
:
80
9
.
20.
Hayano
A
,
Takashima
Y
,
Yamanaka
R
. 
Cell-type-specific sensitivity of bortezomib in the methotrexate-resistant primary central nervous system lymphoma cells
.
Int J Clin Oncol
. 
2019
;
24
:
1020
9
.
21.
Takashima
Y
,
Yoshimura
T
,
Kano
Y
,
Hayano
A
,
Hondoh
H
,
Ikenaka
K
, et al
Differential expression of N-linked oligosaccharides in methotrexate-resistant primary central nervous system lymphoma cells
.
BMC Cancer
. 
2019
;
19
:
910
.
22.
Takashima
Y
,
Terada
M
,
Udono
M
,
Miura
S
,
Yamamoto
J
,
Suzuki
A
. 
Suppression of lethal-7b and miR-125a/b maturation by Lin28b enables maintenance of stem cell properties in hepatoblasts
.
Hepatology
. 
2016
;
64
:
245
60
.
23.
Takashima
Y
,
Sasaki
Y
,
Hayano
A
,
Homma
J
,
Fukai
J
,
Iwadate
Y
, et al
Target amplicon exome-sequencing identifies promising diagnosis and prognostic markers involved in RTK-RAS and PI3K-AKT signaling as central oncopathways in primary central nervous system lymphoma
.
Oncotarget
. 
2018
;
9
:
27471
86
.
24.
Takashima
Y
,
Kawaguchi
A
,
Kanayama
T
,
Hayano
A
,
Yamanaka
R
. 
Correlation between lower balance of Th2 helper T-cells and expression of PD-L1/PD-1 axis genes enables prognostic prediction in patients with glioblastoma
.
Oncotarget
. 
2018
;
9
:
19065
78
.
25.
Takashima
Y
,
Horisawa
K
,
Udono
M
,
Ohkawa
Y
,
Suzuki
A
. 
Prolonged inhibition of hepatocellular carcinoma cell proliferation by combinatorial expression of defined transcription factors
.
Cancer Sci
. 
2018
;
109
:
3543
53
.
26.
Takashima
Y
,
Kawaguchi
A
,
Sato
R
,
Yoshida
K
,
Hayano
A
,
Homma
J
, et al
Differential expression of individual transcript variants of PD-1 and PD-L2 genes on Th-1/Th-2 status is guaranteed for prognosis prediction in PCNSL
.
Sci Rep
. 
2019
;
9
:
10004
.
27.
Garcia-Manteiga
JM
,
Mari
S
,
Godejohann
M
,
Spraul
M
,
Napoli
C
,
Cenci
S
, et al
Metabolomics of B to plasma cell differentiation
.
J Proteome Res
. 
2011
;
10
:
4165
76
.
28.
Shah
HB
,
Koelsch
KA
. 
B-Cell ELISPOT: for the identification of antigen-specific antibody-secreting cells
.
Methods Mol Biol
. 
2015
;
1312
:
419
26
.
29.
Ducker
GS
,
Ghergurovich
JM
,
Mainolfi
N
,
Suri
V
,
Jeong
SK
,
Hsin-Jung Li
S
, et al
Human SHMT inhibitors reveal defective glycine import as a targetable metabolic vulnerability of diffuse large B-cell lymphoma
.
Proc Natl Acad Sci U S A
. 
2017
;
114
:
11404
9
.
30.
Schwarzfischer
P
,
Reinders
J
,
Dettmer
K
,
Kleo
K
,
Dimitrova
L
,
Hummel
M
, et al
Comprehensive metaboproteomics of burkitt's and diffuse large B-cell lymphoma cell lines and primary tumor tissues reveals distinct differences in pyruvate content and metabolism
.
J Proteome Res
. 
2017
;
16
:
1105
20
.
31.
McDonnell
SR
,
Hwang
SR
,
Rolland
D
,
Murga-Zamalloa
C
,
Basrur
V
,
Conlon
KP
, et al
Integrated phosphoproteomic and metabolomic profiling reveals NPM-ALK-mediated phosphorylation of PKM2 and metabolic reprogramming in anaplastic large cell lymphoma
.
Blood
. 
2013
;
122
:
958
68
.
32.
Vander Heiden
MG
. 
Exploiting tumor metabolism: challenges for clinical translation
.
J Clin Invest
. 
2013
;
123
:
3648
51
.
33.
Zhang
J
,
Medina-Cleghorn
D
,
Bernal-Mizrachi
L
,
Bracci
PM
,
Hubbard
A
,
Conde
L
, et al
The potential relevance of the endocannabinoid, 2-arachidonoylglycerol, in diffuse large B-cell lymphoma
.
Oncoscience
. 
2016
;
3
:
31
41
.
34.
Stenson
M
,
Pedersen
A
,
Hasselblom
S
,
Nilsson-Ehle
H
,
Karlsson
BG
,
Pinto
R
, et al
Serum nuclear magnetic resonance-based metabolomics and outcome in diffuse large B-cell lymphoma patients - a pilot study
.
Leuk Lymphoma
. 
2016
;
57
:
1814
22
.
35.
Yang
F
,
Du
J
,
Zhang
H
,
Ruan
G
,
Xiang
J
,
Wang
L
, et al
Serum metabolomics of burkitt lymphoma mouse models
.
PLoS One
. 
2017
;
12
:
e0170896
.
36.
Zhou
QY
,
Wang
YL
,
Li
X
,
Shen
XY
,
Li
KJ
,
Zheng
J
, et al
Metabolomics investigation of cutaneous T cell lymphoma based on UHPLC-QTOF/MS
.
Asian Pac J Cancer Prev
. 
2014
;
15
:
5417
21
.
37.
Le
Y
,
Shen
X
,
Kang
H
,
Wang
Q
,
Li
K
,
Zheng
J
, et al
Accelerated, untargeted metabolomics analysis of cutaneous T-cell lymphoma reveals metabolic shifts in plasma and tumor adjacent skins of xenograft mice
.
J Mass Spectrom
. 
2018
;
53
:
172
82
.
38.
Jiang
G
,
Shen
X
,
Kang
H
,
Li
K
,
Zheng
J
,
Yu
Y
. 
Serum metabolite profiling of cutaneous T-cell lymphoma based on a multiplatform approach
.
J Chromatogr B Analyt Technol Biomed Life Sci
. 
2018
;
1077–1078
:
71
76
.
39.
Warburg
O
. 
On the origin of cancer cells
.
Science
. 
1956
;
123
:
309
14
.