Effect of Glioma Stem Cells-derived Extracellular Vesicles on Glucose Metabolism Reprogramming in Glioma via the miR-26a/KLF4/PI3K/Akt axis


 BackgroundGlioma represents one of the most intractable malignancy occurring in central nervous system. Glioma stem cells (GSCs) constitute a type of seed cells with self-renewal and multi-directional differentiation potentials. This study investigated the mechanism of GSCs-derived extracellular vesicles (EVs) in glucose metabolism reprogramming in glioma.MethodsGlioma cells were treated with different concentrations (10, 20 and 30 μg/mL) of GSCs-EVs. Glucose consumption and lactic acid and adenosine triphosphate (ATP) production were measured and expressions of key enzymes in glycolysis pathway (PFK1, HK2 and PKM2) were detected. The proliferation, invasion and migration of glioma cells were measured. miR-26a expression in GSCs-EVs-treated cells was detected. The targeting relationship between miR-26a and KLF4 was predicted and verified. The phosphorylation levels of PI3K and Akt were detected. Glioma cells were co-incubated with GSCs-EVs and pcDNA-KLF4 to verify the role of PI3K/Akt pathway in glucose metabolism reprogramming.ResultsGlucose consumption and lactic acid and ATP production were promoted, and expressions of PFK1, HK2 and PKM2 were increased in GSCs-EVs-treated cells. GSCs-EVs facilitated proliferation, invasion and migration of glioma cells. miR-26a expression was enhanced in GSCs-EVs-treated cells. GSCs-EVs carried miR-26a into glioma cells and promoted glucose metabolism reprogramming. miR-26a inhibitor reduced the promoting effect of GSCs-EVs on glucose metabolism reprogramming. miR-26a repressed KLF4 expression and activated the PI3K/Akt pathway in glioma cells.ConclusionmiR-26a carried by GSCs-EVs promoted glucose metabolism reprogramming in glioma via targeting KLF4 and activating the PI3K/Akt pathway. This study may offer new insights for targeted intervention of glioma.


Introduction
Glioma represents the most frequent primary malignancies in adults, which may occur in any part of the central nervous system, but predominantly in brain and glial tissues [1]. The management of glioma is confronted with enormous clinical challenges, not only due to their special tumor location, but also because of their malignant biological characteristics, with a high tendency to proliferation, invasion, angiogenesis and metabolic abnormality [2]. The rapid and unrestricted proliferation of tumor cells is a process that consumes energy and resources, and thus, it can be predicted that glucose metabolism will be remarkably altered in tumor progression [3]. There are two main biochemical events in glucose metabolism of tumor cells: (a) increased glucose uptake; (b) aerobic glycolysis, the process of glucose conversing to pyruvate, and eventually producing lactic acid (fermentation) [4]. Just like any other cancers, glioma warrants a sustained source of energy and molecular resources to produce new cells, with a priority given to aerobic glycolysis, which is acknowledged as the Warburg effect [5]. Reprogramming of cellular metabolism is accepted as a shared feature of malignant tumors, and consequently eliminating glycolysis-induced metabolic reprogramming may introduce a potential opportunity for the treatment of glioma [5,6]. Notably, glioma shows extremely metabolic activity and derives energy nearly entirely from glucose [7]. However, the molecular basis and exact mechanism of glucose metabolism reprogramming in glioma remain further clari ed.
Glioma stem cells (GSCs) share biological similarities with normal neural stem cells, but also bear obvious genetic and epigenetic changes [8]. GSCs also adapt to nutritional restriction, hypoxia or chemotherapeutic drug exposure, and interact with microenvironment factors actively to avoid anti-tumor immune response and facilitate tumor angiogenesis and invasion, resulting in tumor recurrence and poor prognosis of patients with glioma [7]. Importantly, it is reported that GSCs can hijack high-a nity glucose uptake to reprogram glucose metabolism, thereby enabling them to survive in a nutritionally dynamic microenvironment [9]. GSCs communicate with tumor microenvironment in multiple models, including the exchange of soluble molecules and extracellular vesicles (EVs), and cell-to-cell contact [10]. EVs are cellderived microparticles that exist in body uids, including microbubbles, exosomes and apoptotic bodies [11]. EVs are implicated in intercellular communication and regulation of tumor microenvironment by mediating the exchange of substances and information between cells [12]. GSCs-EVs contribute to the initiation of glioma by modulating the tumor growth, in ltration and immune invasion [13]. Nevertheless, the role and mechanism of GSCs-EVs in the glucose metabolism reprogramming in glioma is still unclear. This study herein investigated the effects of GSCs-EVs on glucose metabolism reprogramming in glioma, along with its speci c molecular mechanism, which shall open up a promising novel direction for targeted therapy of glioma. The adherent U87 cells cultured in DMEM/F12 containing 10% FBS were prepared into single cell suspension. The cells were resuspended in serum-free medium (2 × 10 5 cells/mL) and cultured in saturated humidity at 37℃ with 5% CO 2 . Then the cells were incubated with GSC markers CD133 (1:300, ab33922, Abcam, Cambridge, MA, USA) and Nestin (1:200, ab105389, Abcam) at 4℃ overnight. After the cell slide was rinsed with phosphate-buffered saline (PBS), the cells were incubated with the secondary antibody immunoglobulin G (IgG) (1:500, ab150088, Abcam) at 37℃ for 1 h. GSCs were screened and isolated using ow cytometry. The differentiation ability of GSCs was evaluated using immuno uorescence. Single cell-derived tumor spheres in serum-free medium were collected and seeded into 6-well plates. The expressions of CD133, glial brillary acidic protein (GFAP) and neuron speci c enolase (NSE) were detected. The expressions of CD133, GFAP and NSE were detected after the cells were cultured in serum medium for 1 week to evaluate the differentiation ability of GSCs.

Materials And
Isolation, culture and identi cation of EVs U87 cells were washed with PBS twice. The medium containing 10% FBS was replaced by 10% EVs-free medium. The cells were cultured for another 48-72 h and then the cell supernatant was collected, followed by the extraction of EVs using differential centrifugation [14]. All the centrifugation steps were completed at 4℃ and the rest steps were performed on ice. The precipitates after centrifugation were resuspended with PBS and centrifuged once again. Finally, the precipitates were resuspended with 50-100 µL PBS and stored at -80℃. The size of EVs was measured by Nanosight [15]. The EV marker proteins CD44 (1:1000, ab189524, Abcam) and CD63 (1:1000, ab134045, Abcam) were identi ed using Western blotting. The EVs were resuspended with 4% paraformaldehyde, and then 5 µL EVs were dripped onto the electron microscope mesh and xed in a dry environment for 20 min. The EVs were washed with PBS and incubated with 1% glutaraldehyde for 5 min. Following washing with distilled water 8 times (2 min/time), EVs were incubated with uranyl oxalic acid solution (pH = 7) for 5 min and hydroxypropyl methylcellulose for 10 min on ice. The excess liquid was removed, and EVs were let stand for 5-10 min and observed under an electron microscope (Zeiss AG, Oberkochen, German). The concentration of EVs was quanti ed by bicinchoninic acid (BCA) assay kit (ab102536, Abcam). EVs at concentration of 10, 20 and 30 µg/mL were used for subsequent experiments.
Detection of the uptake of GSCs-EVs by glioma cells GSCs-EVs were labeled with PKH-26 red uorescent cell linker kit (Sigma-Aldrich, Merck KGaA, Darmstadt, Germany). Brie y, PKH-26-labeled GSCs-EVs were cultured with U87 cells at 37℃ for 12 h. Afterwards, the cells were rinsed with PBS twice, xed with methanol for 20 min and stained with DAPI for 20 s. The stained cells were observed under a uorescence microscope (Olympus, Tokyo, Japan) to determine the uptake of PKH-26-labeled GSCs-EVs by U87 and U251 cells.

Detection of cell proliferation
The cell proliferation was detected using MTT and EdU assay. The cells at passage 3 in logarithmic phase were prepared into single cell suspension (l × 10 5 cells/well). The cell proliferation was detected using the MTT cell proliferation kit (ab211091, Abcam) and Cell-Light™ EdU assay kit (ab219801, Abcam).

Detection of cell invasion and migration
The cell invasion and migration were evaluated by matrix-coated and uncoated 24-well Transwell chamber. Brie y, U87 cells were added into the apical chamber and cultured in serum-free medium containing 10, 20 and 30 µg/mL GSCs-EVs. The complete medium containing 10% FBS was supplemented to the basolateral chamber. After 48 h, the cells in the apical chamber were removed with cotton swabs, and the remaining cells were xed in 4% paraformaldehyde. The cells in the basolateral chamber were removed with a cotton swab, and the migrated cells were xed with methanol and stained with crystal violet. The permeable membrane was observed under a microscope.

Detection of glucose and lactic acid
GSCs were seeded into the 24-well plates (200 µL medium/well). After 24 h, the medium was replaced by fresh medium containing 10, 20 and 30 µg/mL GSCs-EVs, and the cells were incubated for another 48 h. Then, the supernatant and cells were collected. The levels of glucose and lactic acid were determined using the glucose and lactic acid detection kits (Sigma-Aldrich).

Adenosine triphosphate (ATP) detection
GSCs were lysed, and then the lysate was centrifuged at 15000 rpm and 4℃ for 10 min. ATP concentration was measured using ATP detection kit and a microplate Reader (Bio-Rad 680, Bio-Rad, Hercules, CA, USA).

Dual-luciferase reporter gene assay
The binding site between miR-26a and KLF4 was predicted by Targetscan Science & Technology Co., Ltd, Beijing, China) and uorescence intensity was testing using the GLomax20/20 Luminometer (Promega, Madison, WI, USA).

Reverse transcription quantitative polymerase chain reaction (RT-qPCR)
Total RNA was extracted using RNAiso Plus (Takara, Otsu, Shiga, Japan) and TRIzol LS Reagent (Takara). The reliability of extracted RNA was veri ed by formaldehyde denaturation electrophoresis. Then, the extracted RNA was reverse transcribed into cDNA using the PrimeScript™ RT kit (Takara). The relative expression of genes were quanti ed by standard RT-qPCR using SYBR® Premix Ex Taq™ II kit (Takara), with β-actin or U6 as the internal reference. Primer sequences are illustrated in Table 1.

Statistical analysis
Data analysis was introduced using the SPSS 21.0 (IBM Corp., Armonk, NY, USA). Kolmogorov-Smirnov method was adopted to check whether the data were in normal distribution. Data are expressed as mean ± standard deviation. The t test was adopted for comparison between two groups. One-way analysis of variance (ANOVA) was employed for the comparisons among multiple groups, following Tukey's multiple comparisons test. The p < 0.05 meant a statistically signi cance.

Identi cation of GSCs and EVs
GSCs were extracted, identi ed and observed under the light microscope. GSCs grew adherent in the medium. Most of cells showed obvious protuberance and interwove into a net (Fig. 1A). After 48-72 h of incubation, U87 cells grew into round or oval tumor cell spheres, and showed strong refraction and slow growth rate without protuberance (Fig. 1B). More than 95% of GSCs were positive for CD133 and Nestin (Fig. 1C). The differentiated cells presented neuron-or glial cell-like, with the similar morphology to protoglioma cells. After 7 d of differentiation, the astrocyte marker GFAP and neuroendocrine cell marker NSE could be detected (Fig. 1D). It was suggested that GSCs possessed multi-directional differentiation potential. EVs showed cup-shaped under the transmission electron microscope (× 40000) (Fig. 1E). Nanosight results demonstrated that 92.7% of the EVs were located near 120 nm (Fig. 1F). EVs showed high expressions of CD44 and CD63, but no expression of endoplasmic reticulum marker calnexin (Fig. 1G). These results suggested that GSCs and EVs were isolated successfully.

GSCs-EVs promoted glucose metabolism reprogramming in glioma cells
Abnormal glucose metabolism of cancer cells is one of the characteristics of tumor tissue different from normal tissue, mainly manifested as high glucose uptake rate, active glycolysis and high content of lactic acid metabolite [16]. The role and mechanism of GSCs-EVs in glucose metabolism reprogramming in glioma need to be further studied. The effects of different concentrations of GSCs-EVs on glucose consumption and production of lactic acid and ATP in U87 cells were observed. EVs-treated cells exhibited enhanced glucose consumption, as well as elevated lactic acid and ATP production in contrast to the control cells (all p < 0.05; Fig. 2A-C). Meanwhile, the capacity to modulate key enzymes (PFK1, HK2 and PKM2) represent another characteristic of glucose metabolism in tumor cells [17], and key enzymes can affect the glucose metabolism and even the malignant biological behaviors of glioma cells. Hence, the expressions of PFK1, HK2 and PKM2 were detected using RT-qPCR and Western blotting. GSCs-EVs treatment notably increased the mRNA and protein levels of PFK1, HK2 and PKM2 (all p < 0.05; Fig. 2D-G). It was indicated that GSCs-EVs promoted glucose metabolism reprogramming in glioma cells.
GSCs-EVs facilitated the proliferation, invasion and migration of glioma cells U87 cells were incubated with different concentrations of GSCs-EVs and then the cell proliferation was measured using MTT assay. Compared with that of the control cells, the viability of GSCs-EVs-treated cells was signi cantly promoted at 80 h (all p < 0.05; Fig. 3A). The effect of GSCs-EVs on the proliferation of U87 cells was also measured using EdU assay. The number of EdU-positive cells in EVs group was notably higher than that in the control group (all p < 0.05; Fig. 3B). The results of invasion and migration experiments exhibited that EVs treatment promoted the invasion and migration of U87 cells (all p < 0.05; Fig. 3C/D). Taken together, GSCs-EVs facilitated the proliferation, invasion and migration of glioma cells.
GSCs-EVs carried miR-26a into glioma cells GSCs are implicated in cancer initiation and metastasis, which may release EVs and mediate cell communication by carrying miRs [18]. Importantly, miR-26a is closely related to glucose metabolism reprogramming in glioma [19]. We speculated that GSCs-EVs might carry miR-26a into glioma cells. Then miR-26a expression in HA1800, U87 and U251 cells was detected using RT-qPCR. The results demonstrated that miR-26a expression in glioma cells was notably higher than that in normal HA1800 astrocytes (p < 0.01; Fig. 4A). miR-26a expression in cells in the EVs group was higher than that in the GSCs + GW4869 group; no signi cant difference could be observed in miR-26a expression after adding RNA enzyme into GSCs-EVs (Fig. 4B), con rming that miR-26a was encapsulated by GSCs-EVs. After GSCs-EVs treatment, miR-26a expression in U87 and U251 cells was increased signi cantly (p < 0.01; Fig. 4C). The uorescence-labeled GSCs-EVs were co-incubated with U87 and U251 cells, and the uorescence changes were observed after 24 h. The results revealed that GSCs-EVs were internalized by glioma cells (Fig. 4D). These results con rmed that GSCs-EVs carried miR-26a into glioma cells.

GSCs-EVs promoted glucose metabolism reprogramming in glioma cells by carrying miR-26a
To verify that GSCs-EVs played a role in glioma cells by carrying miR-26a, we transfected miR-26a inhibitor into GSCs. miR-26a expression in GSCs was signi cantly reduced after transfection of miR-26a inhibitor (p < 0.01; Fig. 5A). Then, EVs were extracted from miR-26a inhibitor-transfected GSCs and miR-26a expression in EVs was detected. miR-26a expression in EVs was decreased signi cantly after intervention with miR-26a inhibitor (p < 0.01; Fig. 5B). U87 and U251 cells were treated with miR-26a inhibitor-transfected GSCs-EVs, and the results revealed that miR-26a expression in U87 and U251 cells in the EVs + inhibitor group was also reduced (p < 0.01; Fig. 5C). Additionally, the glucose consumption and the lactic acid and ATP production of U87 and U251 cells in the EVs + inhibitor group were notably declined (all p < 0.01; Fig. 5D-F). It was con rmed that GSCs-EVs promoted glucose metabolism reprogramming in glioma cells by carrying miR-26a.

miR-26a carried by GSCs-EVs activated the PI3K/Akt pathway via targeting KLF4
To further study the molecular mechanism of miR-26a in glucose metabolism reprogramming in glioma, we screened the target gene of miR-26a through TargetScan (http://www.targetscan.org/vert_72/) and found that there was a target binding site between 3'-UTR of KLF4 and miR-26a (Fig. 6A). KLF4 is accepted as a tumor suppressor in lung cancer and colorectal cancer [20,21]. The targeting relationship between miR-26a and KLF4 was veri ed using dual-luciferase reporter gene assay (Fig. 6B). RT-qPCR showed that the cells in the miR-26a-mimic group had higher expression of miR-26a and lower expression of KLF4 than that in the miR-26a-NC group (Fig. 6C). Then, KLF4 expression in U87 and U251 cells was detected. EVs-treated cells (10, 20 and 30 µg/mL) showed decreased KLF4 expression compared with the control cells; EVs + inhibitor-treated cells showed increased KLF4 expression compared with the EVstreated cells (all p < 0.01; Fig. 6D/E). These results suggested that miR-26a carried by GSCs-EVs speci cally inhibited KLF4 expression in glioma cells.
miR-26a facilitates angiogenesis of human brain microvascular endothelial cells by activating the PI3K/Akt pathway [18]. Accordingly, we speculated that miR-26a activated the PI3K/Akt pathway via targeting KLF4. The phosphorylation levels of PI3K and Akt in U87 and U251 cells were detected using Western blotting. The phosphorylation levels of p-PI3K and p-Akt were signi cantly promoted in glioma cells treated with different concentrations of EVs (10, 20 and 30 µg/mL), while EVs + inhibitor-treated cells showed decreased phosphorylation levels of PI3K and Akt compared with the EVs-treated cells (all p < 0.01; Fig. 6F). The above results con rmed that miR-26a carried by GSCs-EVs activated the PI3K/Akt pathway via targeting KLF4. miR-26a carried by GSCs-EVs promoted glucose metabolism reprogramming in glioma via targeting KLF4 and activating the PI3K/Akt pathway To verify that miR-26a carried by GSCs-EVs promoted glucose metabolism reprogramming in glioma via targeting KLF4 and activating the PI3K/Akt pathway, we transfected pcDNA-KLF4 and pcNDA-NC plasmids into U87 cells. RT-qPCR and Western blotting con rmed that pcDNA-KLF4 upregulated KLF4 expression in U87 cells (Fig. 7A/B). GSCs-EVs were co-incubated with pcDNA-KLF4-transfected U87 cells (EVs + KLF4 group) and pcNDA-NC-transfected U87 cells (EVs + NC group) respectively. The levels of glucose consumption and lactic acid and ATP production of U87 cells in the EVs + KLF4 group were signi cantly declined compared with those in the EVs + NC group (all p < 0.01; Fig. 7C-E). Additionally, the expressions of PFK1, HK2 and PKM2 were reduced in the EVs + KLF4 group (all p < 0.01; Fig. 7F). Taken together, miR-26a carried by GSCs-EVs promoted glucose metabolism reprogramming in glioma via targeting KLF4 and activating the PI3K/Akt pathway.

Discussion
Reprogramming metabolism is the initial event of glioma tumorigenesis, which dominates the plasticity of tumor cells [7]. GSCs show malignant phenotypes, self-renewal ability and speci c epigenetic characteristics, which possess obvious tumorigenic potential and affect the therapeutic response [22]. We were the rst to reveal that GSCs-EVs facilitated glucose metabolism reprogramming in glioma via the miR-26a/KLF4/PI3K/Akt axis.
Cancer cells prefer aerobic glycolysis to meet the energy demand for rapid growth instead of oxidative phosphorylation, even in the presence of su cient oxygen [23]. Compared with normal cells using mitochondria to oxidize glucose, cancer cells show high glucose consumption and large lactic acid production [24]. GSCs either consume much glucose and generate much lactic acid or consume much oxygen and sustain a high ATP level based on the metabolic features of tumor cells [25]. Additionally, the rapid and continuous production of ATP may constitute another remarkable feature of cancer cell metabolism [26]. We revealed that the glucose consumption, lactic acid and ATP production were notably elevated in EVs-treated cells. The promotion of lactic acid concentration can reconstruct the microenvironment of GSCs, and consequently bene ts the progress of tumor [6]. Glycolytic enzymes are deregulated in cancer cells and implicated in tumorigenesis [27]. Three rate-limiting enzymes including HK2, PFK1 and PKM2, are reported to participate in the aerobic glycolysis [28]. We showed that GSCs-EVs treatment enhanced the mRNA and protein levels of HK2, PFK1 and PKM2. Taken together, GSCs-EVs facilitated glucose metabolism reprogramming in glioma. GSCs possess the tumorigenesis potential and are responsible for tumor proliferation, therapeutic resistance and recurrence [29]. Then, we determined the effect of GSCs-EVs on glioma cells. We exhibited that GSCs-EVs facilitated the proliferation, invasion and migration of glioma cells. Consistently, Li et al. have suggested that glioma-associated human endothelial cell-derived EVs can speci cally enhance the tumourigenicity of GSCs [30] . Brie y, GSCs-EVs facilitated the reprogramming of glucose metabolism in glioma and further promoted the malignant episodes of glioma cells.
EVs carry RNA species (such as lncRNAs, miRs and mRNAs) from donor cells to receptor cells, causing deep phenotypic alterations in tumor microenvironment [31]. miRs are highly conserved non proteincoding RNAs that maintain intracellular homeostasis through negative gene regulation [32]. Notably, miRs are implicated in the modulation of cancer cell metabolism through direct or indirect regulation of genes related to aerobic glycolysis [33]. miR-26a is elevated in glioma tissues and contributes to the initiation of glioma in mice [34]. However, relative little is known about the function of miR-26a in glucose metabolism reprogramming in glioma yet. Consistent with the previous ndings, we showed that miR-26a expression in glioma cells was notably higher than that in normal cells. The glucose consumption and lactic acid and ATP production of glioma cells in the EVs + miR-26a inhibitor group were signi cantly reduced. The elevated miR-26a expression can notably accelerate the growth rate and colony formation of glioma cells, and promote glioma progression and angiogenesis in vivo [19]. miR-26a can accelerate glucose consumption of aerobic glycolysis to meet the energy and biosynthesis of colon cancer cells [35].
Overexpression of miR-26a attenuated HO-induced apoptosis through promoting mitochondrial ATP content and increasing mitochondrial membrane potential [36]. In brief, GSCs-EVs carried miR-26a into glioma cells and promoted reprogramming of glucose metabolism.
To further explore the molecular mechanism of miR-26a affecting glucose metabolism reprogramming in glioma, we screened the target genes of miR-26a through database. miR-26a/KLF4 axis is reported to modulate innate immune signal, macrophage polarization and transport of Mycobacterium tuberculosis to lysosome [37]. The targeting relationship between miR-26a and KLF4 was veri ed using dual-luciferase reporter gene assay in the current study. KLF4 is downregulated in glioma cells, and the enhancement of KLF4 expression contributes to inhibiting cell invasion-mediated epithelial-mesenchymal transition in glioma [38]. KLF4-de cient macrophages show elevated pro-in ammatory levels, promoted bactericidal activity and aberrant metabolism [39]. We showed that KLF4 expression was reduced in EVs-treated glioma cells, and promoted in the EVs + miR-26a-treated cells. In addition, functional rescue experiments were performed. U87 cells transfected with pcNDA-NC or pcDNA-KLF4 were co-incubated with GSCs-EVs. U87 cells in the EVs + KLF4 group had reduced glucose consumption and lactic acid and ATP production, as well as declined expressions of PFK1, HK2 and PKM2. KLF4 activates glycolytic metabolism of breast cancer cells by upregulating platelet isoform of phosphofructokinase, and KLF4 gene knockout inhibits glucose uptake and lactate production [40]. Brie y, miR-26a carried by GSCs-EVs targeted KLF4 in glioma cells.
Thereafter, we shifted to investigating the downstream pathway regulated by miR-26a/KLF4 in glucose metabolism reprogramming in glioma. miR-26a is demonstrated to repress the malignant biological behaviors of papillary thyroid carcinoma by activating the PI3K/Akt pathway [41]. PI3K/Akt pathway exerts different downstream effects on cell metabolism by directly regulating nutrient transporters and metabolic enzymes or by controlling transcription factors that modulate the metabolic pathways [42]. Our results revealed that the phosphorylation levels of p-PI3K and p-Akt were elevated in EVs-treated glioma cells; while cells treated with EVs + miR-26a inhibitor exhibited reduced phosphorylation levels of p-PI3K and p-Akt. Activating the PI3K/Akt signaling pathway is potential to improve glucose metabolism and alleviate insulin resistance in db/db mice [43]. Administration of a PI3K inhibitor impaired glucose uptake [44]. It was con rmed that miR-26a carried by GSCs-EVs activated the PI3K/Akt pathway via targeting KLF4.
Conclusions miR-26a carried by GSCs-EVs promoted the reprogramming of glucose metabolism in glioma via targeting KLF4 and activating the PI3K/Akt pathway. Understanding the mechanism of glucose metabolism reprogramming contributes to determining an appropriate combination of metabolic agents for glioma patients. We also highlight the need for further research to explore the value and importance of inhibiting speci c glycolytic enzymes to reduce tumor burden.  Figure 1 Identi cation of GSCs and EVs. A: U87 cells in the medium were observed under a light microscope; B:

Abbreviations
after 48-72 h of incubation, U87 cells grew into tumor cell spheres; C: expressions of CD133 and Nestin in U87 cells-shaped tumor spheres and adherent cells were measured using immuno uorescence and ow cytometry; D: astrocyte marker GFAP and neuroendocrine cell marker NSE were detected using immuno uorescence; E: morphology of GSCs-EVs was observed under a transmission electron microscope; F: particle size of GSCs-EVs was measured by Nanosight; G: EVs markers CD44, CD63 and calnexin were detected using Western blotting.

Figure 2
GSCs-EVs promoted glucose metabolism reprogramming in glioma cells. A: glucose consumption of U87 and U251 cells was evaluated using glucose detection kit; B: lactic acid production of U87 and U251 cells was evaluated using the lactic acid detection kit; C: ATP production of U87 and U251 cells was evaluated using the ATP detection kit; D-G: mRNA and protein levels of PFK1, HK2 and PKM2 in U87 and U251 cells were detected using RT-qPCR and Western blotting. Glioma cells were treated with different concentrations of GSCs-EVs, and the conditioned medium supplemented with GW4869 was used as the control. Cell experiments were repeated three times independently. Data are presented as mean ± standard deviation and analyzed using one-way ANOVA, followed by Tukey's multiple comparison test, *p < 0.05, **p < 0.01 vs. control group.

Figure 3
GSCs-EVs facilitated the proliferation, invasion and migration of glioma cells. Glioma cells were treated with different concentrations of GSCs-EVs, and the GSCs conditioned medium supplemented with GW4869 was used as the control. A/B: effect of GSCs-EVs on the proliferation of U87 cells was measured using MTT assay and EdU assay; C/D: effect of GSCs-EVs on the invasion and migration of U87 cells was measured using Transwell assays. Cell experiments were repeated three times independently. Data are presented as mean ± standard deviation. Data in panel were analyzed using one-way ANOVA, followed by Tukey's multiple comparison test, *p < 0.05, **p < 0.01 vs. control group.

Figure 4
GSCs-EVs carried miR-26a into glioma cells. A: miR-26a expression in HA1800, U87 and U251 cells was detected using RT-qPCR; B: miR-26a expression in cells treated with GSCs + GW4869 and EVs + RNA enzyme was detected using RT-qPCR; C: miR-26a expression in U87 and U251 cells treated with different concentrations of EVs was detected using RT-qPCR; D: the internalization of PKH-26-labeled GSCs-EVs by U87 and U251 cells was observed under a uorescence microscope. Cell experiments were repeated three times independently. Data are presented as mean ± standard deviation and analyzed using one-way ANOVA, followed by Tukey's multiple comparison test, **p < 0.01.

Figure 5
GSCs-EVs promoted glucose metabolism reprogramming in glioma cells by carrying miR-26a. miR-26a inhibitor was transfected into GSCs with miR-26a NC as the control. EVs were extracted from miR-26a inhibitor-transfected GSCs. miR-26a expression in GSCs, GSCs-EVs and GSCs-EVs-treated U87 and U251 cells was detected using RT-qPCR. A: miR-26a expression in miR-26a inhibitor-transfected GSCs was detected using RT-qPCR; B: miR-26a expression in miR-26a inhibitor-transfected GSCs-EVs was detected using RT-qPCR; C: U87 and U251 cells were treated with miR-26a inhibitor-transfected GSCs-EVs, and then miR-26a expression in U87 and U251 cells was detected using RT-qPCR; D-F: U87 and U251 cells were treated with miR-26a inhibitor-transfected GSCs-EVs, and then the glucose consumption and lactic acid and ATP production in U87 and U251 cells were detected using the kits. Cell experiments were repeated Page 22/23 three times independently. Data are presented as mean ± standard deviation and analyzed using one-way ANOVA, followed by Tukey's multiple comparison test, **p < 0.01 vs. GSCs group or EVs group. Figure 6 miR-26a activated the PI3K/Akt pathway by targeting KLF4. A: target binding site between KLF4 and miR-26a was predicted by TargetScan; B: the targeting relationship between miR-26a and KLF4 was veri ed using dual-luciferase reporter gene assay; C: miR-26a and KLF4 expression in U87 cells in the miR-26amimic group and miR-26a-NC group were detected using RT-qPCR; D-E: KLF4 expression in U87 and U251 cells in each group was detected using RT-qPCR; F: phosphorylation levels of PI3K and Akt in U87 and U251 cells were detected using Western blotting. Cell experiments were repeated three times independently. Data are presented as mean ± standard deviation and analyzed using one-way ANOVA, followed by Tukey's multiple comparison test, **p < 0.01. Figure 7 miR-26a carried by GSCs-EVs promoted glucose metabolism reprogramming in glioma by targeting KLF4 and activating the PI3K/Akt pathway. pcDNA-KLF4 and pcNDA-NC were constructed and transfected into U87 cells. A/B: mRNA and protein levels of KLF4 were detected using Western blotting and RT-qPCR.
GSCs-EVs were co-incubated with pcDNA-KLF4-transfected U87 cells, with pcNDA-NC-transfected U87 cells as the control. C-E: glucose consumption and lactic acid and ATP production of cells were measured using the kits; F: mRNA expression of PFK1, HK2 and PKM2 in cells was detected using RT-qPCR. Cell experiments were repeated three times independently. Data are presented as mean ± standard deviation. Data in panel F were analyzed using one-way ANOVA and data in other panels were analyzed using t test, followed by Tukey's multiple comparison test, **p < 0.01 vs. pcNDA-NC group or EVs + NC group.