Association between Fat-Soluble Vitamin Metabolic Process and Glioma Progression

Abstract

Although multimodality therapy has recently advanced, patients with glioblastoma, one of the most aggressive and deadly types of central nervous system cancer, have a very poor prognosis and rare long-term survival. Vitamins are essential organic nutrients that play a pivotal role in maintaining homeostasis, and various studies have demonstrated the implication of vitamins in the pathophysiology of gliomas. Herein, we aimed to investigate the association of the vitamin metabolic pathway and the corresponding candidate genes for the malignancy, aggressiveness, and poor prognosis of gliomas using The Cancer Genome Atlas database of patients with gliomas. We demonstrated that fat-soluble vitamin metabolic processes are prominently associated with glioma grade, molecular biomarkers, molecular subtypes, and clinical outcomes. Moreover, we identified the key genes related to the fat-soluble metabolic pathway in gliomas using differentially expressed gene analysis. Among them, the expression of the vitamin K epoxide reductase complex subunit 1 (VKORC1), encoding VKOR essential for the vitamin K-dependent γ-carboxylation of target proteins, was prominently associated with not only malignancy, aggressiveness, and poor prognosis of gliomas but also the representative signal pathways related to glioma pathogenesis. Moreover, the inactivation of Vkorc1 by RNA interference decreased the proliferation and migration potential of glioma cells in vitro. Collectively, these findings reveal the pivotal role of fat-soluble vitamin and vitamin K metabolic processes in the pathophysiology of gliomas, thereby identifying a potential target for drug development for the treatment of malignant gliomas.

INTRODUCTION

Gliomas, thought to originate from glial cells, are the most frequent primary brain tumors. They are classified into four grades (grades I, II, III, and IV) based on the degree of malignancy according to WHO criteria: grade I and II tumors are low grades, while grade III and IV tumors are high grades. Glioblastoma (GBM), a grade IV tumor, is an aggressive and deadly type of central nervous system cancer.¹⁾ Isocitrate dehydrogenase (IDH) status is a prognostic biomarker for patients with GBM. Patients with the IDH mutant generally have higher overall survival than those with IDH-wild-type and respond better to temozolomide, an alkylating agent widely used for treating primary and recurrent high-grade gliomas.²⁾ GBM is also classified into mesenchymal, classical, proneural, and neural subtypes.³⁾ Despite recent advances in multimodality therapy with a combination of surgery, radiation therapy, chemotherapy, and molecular-targeted therapy, patients with GBM have a very poor prognosis and rare long-term survival.⁴⁾

Vitamins are essential organic nutrients that play a pivotal role in maintaining homeostasis. Our cells cannot synthesize most of these micronutrients to meet the daily needs of the body; therefore, they must be obtained from the diet.⁵⁾ Vitamins B complex and C are classified as water-soluble, whereas vitamins A, D, E, and K are fat-soluble.⁶⁾ Numerous studies have reported the association between vitamin intake and various types of cancers, including gliomas.⁷⁾ In particular, fat-soluble vitamins have been implicated in the pathophysiology of gliomas because of their ability to access gliomas by penetrating the blood–brain barrier.⁸⁾ However, most studies to date have investigated the in vitro effect of direct vitamin exposure on glioma cell characteristics, such as growth, death, differentiation, and migration, as well as the association between gliomas and dietary vitamin intake or blood vitamin levels. However, few reports have focused on the association between the pathophysiology of gliomas and the metabolic pathways of various types of vitamins.

In this study, we aimed to investigate the implications of vitamin metabolic pathways and their corresponding candidate genes on the malignancy, aggressiveness, and poor prognosis of gliomas. This was accomplished using integrated bioinformatics analyses and data from The Cancer Genome Atlas (TCGA), Repository for Molecular Brain Neoplasia Data (REMBRANDT), and Chinese Glioma Genome Atlas (CGGA) databases of patients with gliomas, accompanied by biological validation using in vitro genetic analyses.

MATERIALS AND METHODS

Preparation of RNA-Sequence Data

Gene expression analysis of patients with glioma was performed with R software (ver. 4.2.2) using the TCGA-GBM, TCGA-GBMLGG, CGGA, and Rembrandt RNA-sequence datasets obtained from the Gliovis data portal (http://gliovis.bioinfo.cnio.es/) (accessed on August 10, 2023).⁹⁾ Gene expression results were normalized by FPKM-UQ, and log2 transformed after adding 0.5 pseudocounts.

Single Sample Gene Set Enrichment Analysis (ssGSEA)

ssGSEA was performed using the GSVA package (ver. 1.46.0). As unnormalized counts data were used in this analysis, the method and kcdf parameters were set to ssGSEA and gaussian, respectively.¹⁰⁾ Six gene sets, namely “GOBP FAT SOLUBLE VITAMIN METABOLIC PROCESS,” “GOBP WATER SOLUBLE VITAMIN METABOLIC PROCESS,” “GOBP VITAMIN A METABOLIC PROCESS,” “GOBP VITAMIN D METABOLIC PROCESS,” “GOBP VITAMIN E METABOLIC PROCESS,” and “GOBP VITAMIN K METABOLIC PROCESS,” included in the GOBP category were obtained from the C5 collection of the Molecular Signatures Database (https://www.gsea-msigdb.org/gsea/msigdb/) (accessed on December 4, 2023).

Survival Analysis

In each gene set, patients were divided into high and low groups based on the median of ssGSEA score or vitamin K epoxide reductase complex subunit 1 (VKORC1) expression, and survival analysis by log-rank test was performed using the survival package (ver. 3, 5–7). Kaplan–Meier curves were constructed using the survminer package (ver. 0.4.9).

Differentially Expressed Genes (DEGs)

Genes with Bonferroni adjusted p < 0.05 by Wilcoxon rank-sum test were defined as DEGs. To identify genes contributing to the prognosis of glioma, 45 genes in the “GOBP FAT SOLUBLE VITAMIN METABOLIC PROCESS” gene set were analyzed. For non-tumor and GBM comparisons, the Rembrandt and TCGA-GBM datasets were used. For Grade II, III, and IV comparisons, the CGGA, Rembrandt, and TCGA-GBMLGG datasets were used. For IDH-wild-type and IDH-mutant comparisons, the CGGA and TCGA-GBM datasets were used.

Gene Set Enrichment Analysis (GSEA)

For GSEA between two groups, the gene lists were sorted in decreasing order based on gene expression fold change. Then, GSEA was performed using the clusterProfiler package (ver. 4.6.2) with the GSEA function set to the following parameters: minGSsize = 5, maxGSsize = 2775, and pvalueCutoff = 1.00. The HALLMARK gene sets were collected from the H category of the Molecular Signatures Database (https://www.gsea-msigdb.org/gsea/msigdb/) (accessed on April 25, 2024) using the msigdbr function in the msigdbr (ver 7.5.1) package.

Lentiviral Supernatant Preparation and Infection

HEK293T cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and antibiotics (100 U/mL penicillin and 100 µg/mL streptomycin) at 37 °C in a 5% CO₂ incubator. HEK293T cells were transfected with lentiviral, packaging, and envelope plasmids using the calcium phosphate method to produce lentiviral particles. Fourteen hours after transfection, the medium was replaced with fresh DMEM containing 10% FBS. Forty-eight hours after transfection, lentiviral supernatant was collected and the viral supernatant was infected with GL261 cells for 24 h.¹¹⁾ shVkorc1#1 (TRCN0000042152) and shVkorc1#2 (TRCN0000042151) were purchased from Sigma-Aldrich (St. Louis, MO, U.S.A.); pLKO.1 puro plasmid (#8453) was obtained from Addgene (Watertown, MA, U.S.A.).

Real-Time Quantitative PCR (RT-qPCR)

Total RNA was extracted using the FastGene RNA Basic Kit (Nippon Genetics, Tokyo, Japan) according to the manufacturer’s instructions. cDNA was reverse-transcribed from 2.5 µg of the total RNA using M-MLV Reverse Transcriptase (Invitrogen). RT-qPCR was performed on an MX3000P (Agilent Technologies (St. Clara, CA, U.S.A.)) using THUNDERBIRD^® NEXT SYBR^® qPCR Mix (TOYOBO, Tokyo, Japan).¹²⁾ The primer sequences were as follows: Vkorc1-F (5′-CCTCAACCAATCCAACAGCA-3′), Vkorc1-R (5′-GGACACCAGGGAACTCAGCA-3′), Actb-F (5′-TCTCCCTCACGCCATCCT-3′), and Actb-R (5′-TCACGCACGATTTCCCTCT-3′).

3-(4,5-Dimethylthiazol-2-yl)-2, 5-Diphenyltetrazolium Bromide (MTT), Apoptosis, and Wound Healing Assays

MTT, wound healing, and apoptosis assays were performed to evaluate cell properties.^13,14) For the MTT assay, GL261 cells were seeded in a 48 well plate at a density of 2000 cells/well. On each measurement day, 200 µL of 0.5 mg/mL MTT solution in phosphate buffered saline (PBS) was added to each well. After 4 h of incubation, 200 µL of 0.04 mol/L HCl in isopropanol was added to dissolve the formazan crystals. The absorbance of the samples was measured at 550 nm using a microplate reader. For the wound healing assay, wounds were formed using sterile 100 µL tips, and images were taken at 0 and 24 h using a BZ-X800 Analyzer (KEYENCE, Osaka, Japan). The wound closure area:wound area ratio (migration rate) was calculated using ImageJ software. For the apoptosis assay, cells were incubated with PE-labeled Annexin V (1 : 50, #560930, BD, Franklin Lakes, NJ, U.S.A.) and 7-AAD (1 : 1000, #559925, BD) for 30 min at 4 °C in the dark, and then analyzed using a CytoFLEX S flow cytometer (Beckman, Brea, CA, U.S.A.).

Statistical Analysis

All results are expressed as mean ± standard error (S.E.). Statistical significance was determined using the two-tailed Student’s t-test, one-way ANOVA with Dunnett’s post hoc test, or two-way ANOVA with Tukey’s HSD test.

RESULTS

Vitamin Metabolic Process Is Linked to the Malignancy, Aggressiveness, and Poor Prognosis of Gliomas

First, we performed ssGSEA to explore the alternation of vitamin metabolic pathway in gliomas using TCGA database and observed that the gene sets of both “fat-soluble vitamin metabolic process” and “water-soluble vitamin metabolic process” were significantly enriched in GBM tissues compared to non-tumor brain tissues (Fig. 1A), in patients with grade IV compared to patients with grade II and III tumors (Fig. 1B), in patients exhibiting IDH-wild-type status compared to patients harboring IDH-mutant (Fig. 1C), and in patients with the mesenchymal subtype, the most aggressive among the molecular GBM subtypes, compared to patients with classical and proneural subtypes (Fig. 1D). Next, we divided the patients with glioma into two groups, those with high or low vitamin metabolic process, as determined by their ssGSEA score based on the enrichment of gene sets, and assessed whether the vitamin metabolic pathway was associated with poor prognosis in these patients. Using TGCA database, survival analysis demonstrated that patients with higher enrichment of “fat-soluble vitamin metabolic process” and “water-soluble vitamin metabolic process” exhibited significantly shorter overall survival times than those with lower enrichment (Fig. 1E).

Fig. 1. Vitamin Metabolic Process Gene Sets Are Enriched in Gliomas and Correlate with Poor Prognosis

(A–D) Comparison of ssGSEA scores for vitamin metabolic process gene sets. (A) Non-tumor (n = 10) and GBM (n = 528) in TCGA-GBM cohort (** p < 0.01, *** p < 0.001). (B) Grade II (n = 226), grade III (n = 244), and grade IV (n = 150) in TCGA-GBMLGG cohort (*** p < 0.001, n.s.: not significant). (C) IDH-wildtype (n = 372) and IDH-mutant (n = 30) in TCGA-GBM cohort (** p < 0.01, *** p < 0.001). (D) Classical (n = 199), mesenchymal (n = 166), and proneural (n = 163) subtypes in TCGA-GBM cohort (*** p < 0.001, n.s.: not significant). (E) Kaplan–Meier curves in high (n = 333) and low (n = 334) patients for vitamin metabolic process.

Collectively, these results suggest that the vitamin metabolic pathway is associated with malignancy, aggressiveness, and survival outcomes in patients with glioma.

Fat-Soluble Vitamin Metabolic Process Is Linked to the Malignancy, Aggressiveness, and Poor Prognosis of Gliomas

Next, we performed ssGSEA using TCGA database to investigate which vitamin metabolic process, among the fat-soluble vitamins, is associated with glioma pathogenesis. The gene sets involved in the metabolism of vitamins D and E were significantly upregulated in GBM tissues compared to non-tumor brain tissues (Fig. 2A). Gene sets related to vitamin K metabolic processes were positively associated with increased glioma grade (grades II, III, and IV), while gene sets related to the metabolic processes of vitamins A, D, and E were significantly enriched in grade IV compared to grades II and III (Fig. 2B). Moreover, gene sets involved in the metabolic processes of vitamins A, D, E, and K were significantly upregulated in patients exhibiting IDH-wild-type status compared to those harboring IDH mutations (Fig. 2C). Gene sets involved in the metabolic processes of vitamins D, E, and K were significantly higher in patients with the mesenchymal subtype than in those with the classical and proneural subtypes (Fig. 2D). Moreover, Kaplan–Meier survival analysis demonstrated that patients with higher enrichment of gene sets related to the metabolic processes of vitamins A, D, E, and K exhibited significantly shorter overall survival times than those with lower enrichment (Fig. 2E).

Fig. 2. Fat-Soluble Vitamin Metabolic Process Gene Sets Are Enriched in Gliomas and Correlate with Poor Prognosis (A–D) Comparison of ssGSEA Scores for Fat-Soluble Vitamin Metabolic Process Gene Sets

(A) Non-tumor (n = 10) and GBM (n = 528) in TCGA-GBM cohort (* p < 0.05, *** p < 0.001, n.s.: not significant). (B) Grade II (n = 226), grade III (n = 244), and grade IV (n = 150) in TCGA-GBMLGG cohort (* p < 0.05, *** p < 0.001, n.s.: not significant). (C) IDH-wildtype (n = 372) and IDH-mutant (n = 30) in TCGA-GBM cohort (** p < 0.01, *** p < 0.001). (D) Classical (n = 199), mesenchymal (n = 166), and proneural subtypes (n = 163) in TCGA-GBMLGG cohort (* p < 0.05, *** p < 0.001, n.s.: not significant). (E) Kaplan–Meier curves in high (n = 333) and low (n = 334) patients for vitamin metabolic process.

Collectively, these results suggest that the metabolic pathways of all four fat-soluble vitamins (A, D, E, and K) are associated with malignancy, aggressiveness, and survival outcomes in patients with gliomas.

VKORC1 Expression Is Associated with the Malignancy, Aggressiveness, and Poor Prognosis of Gliomas

We then screened DEGs related to fat-soluble metabolic pathways in gliomas using TCGA, REMBRANDT, and CGGA databases. As shown in Fig. 3A, 11 overlapping genes were co-upregulated in GBM tissues compared to non-tumor brain tissues in TCGA and REMBRANDT databases, 10 overlapping genes were co-upregulated in patients with GBM harboring IDH-wild-type compared to those with mutant status in TCGA and CGGA databases, and 15 overlapping genes were co-upregulated in grade IV compared to grades II and III in TCGA, CGGA, and REMBRANDT databases (Supplementary Table 1). Finally, among these, 6 overlapping potential genes were identified, namely CYP family 27 subfamily A member 1 (CYP27A1), phospholipid transfer protein (PLTP), retinol-binding protein 1 (RBP1), snail family transcriptional repressor 2 (SNAI2), nuclear factor kappa B subunit 1 (NFKB1), and VKORC1 (Fig. 3A).

Fig. 3. VKORC1 Expression Is Increased in Gliomas and Correlates with Poor Prognosis

(A) Venn diagram showing DEGs by histology (GBM/Non-tumor >1.0), IDH status (Mutant/Wildtype <1.0), and grade (Grade IV/Grades II, III >1.0). (B–E) Comparison of VKORC1 expression. (B) Non-tumor (n = 10) and GBM (n = 528) in TCGA-GBM cohort (*** p < 0.001). (C) Grade II (n = 226), grade III (n = 244), and grade IV (n = 150) in TCGA-GBMLGG cohort (*** p < 0.001). (D) IDH-wildtype (n = 372) and IDH-mutant (n = 30) in TCGA-GBM cohort (** p < 0.01). (E) Classical (n = 199), mesenchymal (n = 166), and proneural (n = 163) subtypes in TCGA-GBM cohort (* p < 0.05, *** p < 0.001). (F) Kaplan–Meier curves in VKORC1 high expression patients (n = 333) and low expression patients (n = 334). (G) A schematic diagram and distribution chart of TCGA-GBM cohort divided into VKORC1 high expression (n = 264) and low expression groups (n = 264) for GSEA. (H) GSEA of the HALLMARK gene sets comparing VKORC1 high expression patients (n = 333) and low expression patients (n = 334).

According to the data from TCGA database, the expression of VKORC1 was significantly upregulated in GBM tissues (Fig. 3B), was associated with increased glioma grade (Fig. 3C), was significantly upregulated in patients with IDH-wild-type status (Fig. 3D), and was the highest in the mesenchymal subtype (Fig. 3E). Kaplan–Meier survival analysis revealed that increased expression of VKORC1 was significantly associated with poor prognosis (Fig. 3F). We then divided the patients into VKORC1^high and VKORC1^low groups based on VKORC1 expression to explore the associated signal pathways in gliomas (Fig. 3G). A significant enrichment for gene sets related to the pathophysiology of gliomas such as “epithelial mesenchymal transition,” “Glycolysis,” “MYC targets,” “Hypoxia,” and “MTORC1” was observed in VKORC1^high patients (Fig. 3H).

Taken together, these findings suggest that VKORC1 expression is associated with-malignancy, aggressiveness, and survival outcomes in patients with gliomas.

Disrupting Vkorc1 Decreases Proliferation and Migration of Glioma Cells in Vitro

To validate the results of our bioinformatics analyses, we elucidated the functional significance of Vkorc1 in glioma cells in vitro by targeting Vkorc1 expression using lentiviral shRNA in GL261 glioma cells (Fig. 4A). Vkorc1 mRNA levels were significantly reduced by shVkorc1 in GL261 glioma cells compared to controls (Fig. 4B). Vkorc1 silencing significantly reduced the proliferation and migration potential of glioma cells, as determined by MTT and wound healing assays, respectively (Figs. 4C, 4D). On the contrary, targeting Vkorc1 did not significantly alter cell apoptosis, as determined by Annexin V staining (Fig. 4E).

Fig. 4. Inhibition of Vkorc1 Suppresses Proliferation and Migration of Glioma Cells

(A) GL261 cells were infected with shVkorc1, followed by MTT, wound healing, and apoptosis assays. (B) Vkorc1 knockdown was verified via RT-qPCR (n = 4, *** p < 0.001). (C) Cell viability (n = 5, *** p < 0.001), (D) migration ability (n = 3, ** p < 0.01), and (E) cell apoptosis (n = 4, n.s.: not significant) of GL261 cells were analyzed following Vkorc1 knockdown. The representative images for wound healing and apoptosis assays are presented in D (scale bar = 200 µm) and E, respectively.

Collectively, these results indicate that Vkorc1 could be implicated in the regulation of glioma cell properties in vitro.

DISCUSSION

Post-translational protein modifications are frequently observed in gliomas, and we had recently demonstrated the implications of phosphorylation, ubiquitination, and glycosylation of specific proteins in the pathogenesis of glioma malignancy.^15–17) Vitamin K is an isoprenoid quinone and an essential fat-soluble micronutrient. Phylloquinone (vitamin K1) and menaquinones (vitamin K2) act as cofactors for the γ-carboxylation of target proteins, a rare post-translational protein modification.¹⁸⁾ Enzymes involved in vitamin K-dependent γ-carboxylation include gamma-glutamyl carboxylase (GGCX) and VKOR. During γ-carboxylation by vitamin K, GGCX converts specific glutamate (Glu) residues to γ-carboxyglutamate (Gla) residues in target proteins, and VKOR reduces vitamin K epoxides to hydroquinones to sustain the γ-carboxylation by GGCX.¹⁹⁾ Gla proteins contribute to the pathophysiology of gliomas in vitro and in vivo. Protein S, a Gla protein that activates TAM (TYRO3, AXL, and MERTK) receptor tyrosine kinases, is overexpressed in GBM and regulates the proliferation, migration, and invasion of glioma cells.²⁰⁾ Matrix Gla protein, a small secretory protein containing five Gla residues, is also overexpressed in GBM and regulates various characteristics of glioma cells.²¹⁾ However, the roles of the vitamin K cycle enzymes GGCX and VKOR in the pathophysiology of gliomas have not been extensively investigated. Given that VKORC1L1, a paralog of VKORC1, protects cancer cells from ferroptosis by generating the reduced form of vitamin K, we should further investigate the possible mechanisms by which vitamin K metabolism contributes to the pathogenesis of glioma.²²⁾ Although in vivo analyses are required to further validate our results, to the best of our knowledge, this is the first study using integrated bioinformatics analyses and in vitro genetic analyses to demonstrate that VKORC1 is strongly associated with the malignancy, aggressiveness, and poor prognosis of gliomas.

Notably, expression analysis of DEGs linked to fat-soluble vitamin metabolic pathways revealed the possible involvement of alternative candidate genes in the pathophysiology of gliomas. Among the DEGs, PLTP plays an important role in plasma lipoprotein metabolism; regulates the bioavailability of vitamin E; is overexpressed in GBM; and controls the growth, migration, and apoptosis of glioma cells.²³⁾ RBP1 is essential for vitamin A stability and metabolism, and promotes the malignancy of non-glioblastomatous diffuse gliomas.²⁴⁾ CYP27A1, a mitochondrial enzyme belonging to the CYP family, hydroxylates vitamin D3 and is upregulated in GBM tissues.²⁵⁾ Meanwhile, the implication of VKOR, an essential enzyme for vitamin K cycle for γ-carboxylation of targeted proteins, on the pathophysiology of gliomas has not been reported thus far. Despite the need for further exploration of alternative candidate genes, we revealed the crucial role of the fat-soluble vitamin metabolism pathway and VKORC1 in the regulation of the malignancy, aggressiveness, and poor prognosis of gliomas. Our findings suggest that the fat-soluble vitamin metabolism pathway and VKORC1 could represent novel targets for drug development in the treatment of malignant gliomas.

Acknowledgments

This work was supported in part by the Japan Society for the Promotion of Science (20H03407 to E.H.). The results shown here are in part based upon data generated by the TCGA Research Network: https://www.cancer.gov/tcga.

Conflict of Interest

The authors declare no conflict of interest.

Supplementary Materials

This article contains supplementary materials.

REFERENCES

• 1) Louis DN, Perry A, Wesseling P, Brat DJ, Cree IA, Figarella-Branger D, Hawkins C, Ng HK, Pfister SM, Reifenberger G, Soffietti R, von Deimling A, Ellison DW. The 2021 WHO Classification of Tumors of the Central Nervous System: a summary. Neuro-oncol., 23, 1231–1251 (2021).
• 2) Melhem JM, Detsky J, Lim-Fat MJ, Perry JR. Updates in IDH-wildtype glioblastoma. Neurotherapeutics, 19, 1705–1723 (2022).
• 3) Behnan J, Finocchiaro G, Hanna G. The landscape of the mesenchymal signature in brain tumours. Brain, 142, 847–866 (2019).
• 4) van Solinge TS, Nieland L, Chiocca EA, Broekman MLD. Advances in local therapy for glioblastoma—taking the fight to the tumour. Nat. Rev. Neurol., 18, 221–236 (2022).
• 5) Semba RD. The discovery of the vitamins. Int. J. Vitam. Nutr. Res., 82, 310–315 (2012).
• 6) Nur SM, Rath S, Ahmad V, Ahmad A, Ateeq B, Khan MI. Nutritive vitamins as epidrugs. Crit. Rev. Food Sci. Nutr., 61, 1–13 (2021).
• 7) Mamede AC, Tavares SD, Abrantes AM, Trindade J, Maia JM, Botelho MF. The role of vitamins in cancer. a review. Nutr. Cancer, 63, 479–494 (2011).
• 8) Youness RA, Dawoud A, ElTahtawy O, Farag MA. Fat-soluble vitamins: updated review of their role and orchestration in human nutrition throughout life cycle with sex differences. Nutr. Metab. (Lond.), 19, 60 (2022).
• 9) Bowman RL, Wang Q, Carro A, Verhaak RG, Squatrito M. GlioVis data portal for visualization and analysis of brain tumor expression datasets. Neuro-oncol., 19, 139–141 (2017).
• 10) Hänzelmann S, Castelo R, Guinney J. GSVA: gene set variation analysis for microarray and RNA-Seq data. BMC Bioinformatics, 14, 7 (2013).
• 11) Ozaki K, Yamada T, Horie T, Ishizaki A, Hiraiwa M, Iezaki T, Park G, Fukasawa K, Kamada H, Tokumura K, Motono M, Kaneda K, Ogawa K, Ochi H, Sato S, Kobayashi Y, Shi YB, Taylor PM, Hinoi E. The L-type amino acid transporter LAT1 inhibits osteoclastogenesis and maintains bone homeostasis through the mTORC1 pathway. Sci. Signal., 12, eaaw3921 (2019).
• 12) Nakamura Y, Hinoi E, Iezaki T, Takada S, Hashizume S, Takahata Y, Tsuruta E, Takahashi S, Yoneda Y. Repression of adipogenesis through promotion of Wnt/β-catenin signaling by TIS7 up-regulated in adipocytes under hypoxia. Biochim. Biophys. Acta Mol. Basis Dis., 1832, 1117–1128 (2013).
• 13) Fukasawa K, Park G, Iezaki T, Horie T, Kanayama T, Ozaki K, Onishi Y, Takahata Y, Yoneda Y, Takarada T, Kitajima S, Vacher J, Hinoi E. ATF3 controls proliferation of osteoclast precursor and bone remodeling. Sci. Rep., 6, 30918 (2016).
• 14) Osumi R, Sugihara K, Yoshimoto M, Tokumura K, Tanaka Y, Hinoi E. Role of proteoglycan synthesis genes in osteosarcoma stem cells. Front. Oncol., 14, 1325794 (2024).
• 15) Hiraiwa M, Fukasawa K, Iezaki T, Sabit H, Horie T, Tokumura K, Iwahashi S, Murata M, Kobayashi M, Suzuki A, Park G, Kaneda K, Todo T, Hirao A, Nakada M, Hinoi E. SMURF2 phosphorylation at Thr249 modifies glioma stemness and tumorigenicity by regulating TGF-β receptor stability. Commun. Biol., 5, 22 (2022).
• 16) Fukasawa K, Lyu J, Kubo T, Tanaka Y, Suzuki A, Horie T, Tomizawa A, Osumi R, Iwahashi S, Tokumura K, Murata M, Kobayashi M, Todo T, Hirao A, Hinoi E. MEK5-ERK5 axis promotes self-renewal and tumorigenicity of glioma stem cells. Cancer Res. Commun., 3, 148–159 (2023).
• 17) Tokumura K, Sadamori K, Yoshimoto M, Tomizawa A, Tanaka Y, Fukasawa K, Hinoi E. The bioinformatics identification of potential protein glycosylation genes associated with a glioma stem cell signature. BioMedInformatics, 4, 75–88 (2024).
• 18) Welsh J, Bak MJ, Narvaez CJ. New insights into vitamin K biology with relevance to cancer. Trends Mol. Med., 28, 864–881 (2022).
• 19) Shearer MJ, Okano T. Key pathways and regulators of vitamin K function and intermediary metabolism. Annu. Rev. Nutr., 38, 127–151 (2018).
• 20) Ginisty A, Oliver L, Arnault P, Vallette F, Benzakour O, Coronas V. The vitamin K-dependent factor, protein S, regulates brain neural stem cell migration and phagocytic activities towards glioma cells. Eur. J. Pharmacol., 855, 30–39 (2019).
• 21) Fu M-H, Wang C-Y, Hsieh Y-T, Fang K-M, Tzeng S-F. Functional role of matrix gla protein in glioma cell migration. Mol. Neurobiol., 55, 4624–4636 (2018).
• 22) Yang X, Wang Z, Zandkarimi F, Liu Y, Duan S, Li Z, Kon N, Zhang Z, Jiang X, Stockwell BR, Gu W. Regulation of VKORC1L1 is critical for p53-mediated tumor suppression through vitamin K metabolism. Cell Metab., 35, 1474–1490.e8 (2023).
• 23) Dong W, Gong H, Zhang G, Vuletic S, Albers J, Zhang J, Liang H, Sui Y, Zheng J. Lipoprotein lipase and phospholipid transfer protein overexpression in human glioma cells and their effect on cell growth, apoptosis, and migration. Acta Biochim. Biophys. Sin. (Shanghai), 49, 62–73 (2017).
• 24) Wu W, Wang Y, Niu C, Wahafu A, Huo L, Guo X, Xiang J, Li X, Xie W, Bai X, Wang M, Wang J. Retinol binding protein 1-dependent activation of NF-κB signaling enhances the malignancy of non-glioblastomatous diffuse gliomas. Cancer Sci., 113, 517–528 (2022).
• 25) Xin J, Zheng LM, Sun DK, Li XF, Xu P, Tian LQ. miR-204 functions as a tumor suppressor gene, at least partly by suppressing CYP27A1 in glioblastoma. Oncol. Lett., 16, 1439–1448 (2018).