Introduction

Glioblastoma multiforme (GBM) is the most serious brain cancer originating from astrocytes, classified as WHO grade IV1,2, and lacks medicine that shows noticeable effects2,3,4. Due to GBM's exceptional invasion capabilities, surgical excision alone will not allow for full therapy of the disease as it has already spread to other regions and will almost certainly reoccur5. This enormous invasive capacity causes functional and structural destruction of the normal brain6.

Unlike metastasis, which spreads to other regions through the circulatory or lymphatic systems, invasion is the penetration and spread of malignancy from the disease site to surrounding tissues7. Cancer cell invasion is caused by a wide range of parameters, such as molecular signals, cellular junctions, and interactions with the matrix8,9, however, extracellular matrix (ECM), Ca2+ & glutamate signaling are the main mechanisms in GBM6. GBM upregulates the expression of various signaling molecules such as G protein-coupled receptor (GPCR) and receptor tyrosine kinase (RTK), and this signaling causes calcium responses in cells and endoplasmic reticulum (ER)6. Calcium flows in from the outside through store-operated channels (SOCs), Ca2+-permeable transient receptor potential (TRP) channels, and voltage-gated calcium channels (VGCCs) allowing the invasion and migration of GBM6,10. GBM obtains glutamate which plays a critical role in cell invasion through autocrine and paracrine ways. SLC7A11 (or xCT), a glutamate/cysteine antiporter, is highly expressed in GBM and participates in autocrine signaling by exporting glutamate11,12. Further, invasion is promoted by astrocytes and neuronal paracrine glutamate surrounding the GBM13,14. Glutamate from the same or neighboring cells activates AMPA receptors (AMPARs), allowing Ca2+ to flow into the GBM, which is proposed to be important for cell mobility15.

Metabolic dysregulation is one of the hallmarks of cancer and is important in promoting the development of migration, proliferation, and invasion16,17. Likewise, the integrity of GBM largely depends upon the energy metabolism process which includes glycolysis, TCA cycle, and mitochondrial bioenergetics, as well as mitochondria homeostasis18. The cell viability and the growth of GBM cells were severely affected when the biosynthesis of several metabolites in glycolysis and TCA cycles was highly impaired18. Like the majority of cancers, GBM has an uncommon aerobic glycolysis state that GBM preferably utilizes an aerobic glycolytic pathway rather than oxidative phosphorylation resulting in the production of lactate, known as the Warburg effect19,20. It is still unknown, however, why the aberrant energy metabolism occurs in GBM and how it affects its special functions including invasion and proliferation of GBM.

NAD+ (nicotinamide adenine dinucleotide) and NADH (reduced form of NAD+) are crucial redox cofactors in diverse fields of metabolism and the balance of NADH/NAD+ ratio is essential for sustaining a wide range of biological activities due to their critical functions in controlling the cellular redox condition, energy metabolism, mitochondrial function, cellular homeostasis, and signaling pathways21,22,23,24,25,26,27. Disturbances in the redox equilibrium of NADH/NAD+ are associated with a wide range of clinical disorders, including cancer, cardiovascular illnesses, and neurological diseases, and thus NAD+/NADH metabolic pathway has been a target for new drug development20,23,28,29,30. In particular, cancer cells have higher ratios of NAD+/NADH than non-cancerous cells, indicating the critical function that NAD+ plays in this metabolic process in cancer cells27,31. In 2013, it was reported that NAD(H) level is an important metabolic component in glioma cells and regulates glioma cell mobility30. This study suggested that when the function of NAMPT (nicotinamide phosphoribosyltransferase) is blocked through genetic or pharmacological inhibition, NAD+ is depleted in cancer cells, resulting in reduced mobility and proliferation. However, it remains uncertain how physiological fluctuations in the NAD+/NADH ratio, rather than absolute NAD(H) levels, impact GBM function.

The ratio change of NAD+/NADH can be controlled by GEMMs (Genetically Encoded Tools for Manipulation of Metabolisms), which can directly manipulate specific metabolite levels and ratios32. GEMMs can be used in a spatiotemporal manner and do not affect anything other than the level or ratio of specific metabolites such as ATP/ADP, NAD+/NADH, and CoQ/CoQH32. LbNOX, one of the GEMMs, an NADH Oxidase obtained from Lactobacillus brevis, is a specific enzyme that alters the NAD+/NADH ratio by converting NAD+ from NADH33. Using LbNOX, studies have shown that the ratio of NAD+/NADH plays important biological roles in cells. Proliferative and metabolic defects brought on by a compromised ETC were lessened when LbNOX was expressed in the cytosol or mitochondria in HeLa cells34. Also, hepatic LbNOX alleviated hepatic insulin resistance during hyperinsulinaemic–euglycaemic clamps and enhanced glucose tolerance in HFD mice25. As such, whereas the significance of the NAD+/NADH ratio has been established in normally growing cells, little is known about it in the metabolically aberrant GBM.

In this study, we used the U-87 MG GBM cell line and revealed that the increased ratio of NAD+/NADH by LbNOX and mitoLbNOX (mitochondrial targeting LbNOX) plays a significant biological role in GBM. Unlike previously reported papers30, in this study, we used an enzyme that acts in a spatiotemporal manner, called LbNOX. The experiment was conducted specifically for cytoplasm and mitochondria, and the mitoLbNOX group showed a much more positive phenotype. Through metabolic profiling, we showed that cells expressing LbNOX had an elevated accumulation of glutamate. Subsequently, we performed glutamate imaging and verified that the quantity of glutamate released in an activity-dependent manner was diminished. In addition, LbNOX and mitoLbNOX expressing GBM exhibit a significant reduction in migration, proliferation, and invasion. To identify the mechanism of LbNOX-mediated accumulation of glutamate in GBM, calcium imaging was performed to investigate whether LbNOX/mitoLbNOX calcium signaling is involved in the glutamate release, and we found that activity-dependent calcium release was decreased. Taken together, we conclude that the increase in the NAD+/NADH ratio regulates GBM invasion by affecting calcium signaling, which subsequently results in the alteration of intracellular glutamate signaling.

Results

The NAD+/NADH ratio in U-87 MG was increased by LbNOX or mitoLbNOX

To determine the role of the NAD+/NADH ratio, we first expressed LbNOX and mitoLbNOX in the U-87 MG GBM cell line. LbNOX and mitoLbNOX sequences were cloned into pHR-CMV-TetO2_3C-Avi-His6__IRES-EmGFP vector (Fig. 1A). The construct was transfected with packaging and envelope vector into HEK 293T cells to create lentivirus (Fig. 1A), and lentivirus was infected into U-87 MG to express LbNOX or mitoLbNOX (Fig. 1A). The visualization of LbNOX and mitoLbNOX expression in U-87 MG cells was achieved through the detection of green fluorescent protein (GFP) fluorescence (Fig. 1A and B).

Fig. 1
figure 1

Effect of LbNOX, mitoLbNOX expression on NAD+/NADH ratio and ATP levels in U-87 MG. (A) The map of the lentiviral CMV-LbNOX and CMV-mitoLbNOX constructs, as well as a step-by-step procedure for virus collection and subsequent infection of U-87 MG. (B) The representative photographs of LbNOX (left) and mitoLbNOX (right) in U-87 MG, with photographs of GFP and DIC. Scale bar = 20 μm. (C) The NAD+/NADH ratio curve in each condition by time using kinetic mode measurement. Each plot contains 3 wells of data. (D) The bar graph shows the concentration of ATP in each condition (naive and mitoLbNOX: n = 6, LbNOX: n = 5). All individual dots are data from a single well. All data are reported as mean with SEM. All data were analyzed by one-way ANOVA followed by a Tukey’s multiple comparisons test. **P < 0.01, *P < 0.05.

Full size image

We measured the NAD+/NADH ratio to corroborate any alterations in the NAD+/NADH ratio resulting from the expression of LbNOX and mitoLbNOX. The data was acquired in kinetic mode utilizing a spectrophotometer, revealing a consistent elevation in the NAD+/NADH ratio at all examined time points due to the presence of LbNOX and mitoLbNOX (Mean value of ratio, LbNOX: 5.71, mitoLbNOX: 4.47, naive: 3.67) (Fig. 1C). When compared to the unaltered U-87 MG cell line, those expressing mitoLbNOX exhibited a discernible rise in the NAD+/NADH ratio, albeit without statistical significance (Fig. 1C). This was consistent with previous research results35. An ATP assay was conducted to investigate whether the observed elevation in the NAD+/NADH ratio had any impact on ATP production within the U-87 MG cell line. Our results demonstrated no statistically significant differences in intracellular ATP levels across all conditions. The mean ATP concentrations (in mM) for the naive, LbNOX, and mitoLbNOX groups were 0.77, 0.83, and 0.80, respectively (Fig. 1D). These findings suggest that the expression of LbNOX or mitoLbNOX in U-87 MG is not involved in ATP production but rather specifically regulates the NAD+/NADH ratio.

LbNOX altered glutamate signaling in U-87 MG

Metabolic profiling was performed to determine the impact of LbNOX expression and an increased NAD+/NADH ratio on the metabolism of GBM. PCA analysis shows the two samples consisting of significantly changed 398 metabolites (P < 0.05), exhibited distinct differences between the naive and LbNOX-expressing U-87 MG groups (Fig. 2A and Supplementary Table 1). Figure 2B presents a heatmap displaying the relative abundance of selected metabolites in U-87 MG LbNOX cells compared to U-87 MG naive cells. This heatmap represents the top 30 metabolites that showed an increase compared to the naive state upon LbNOX expression. Among these, metabolites associated with glutamate, caffeine, sphingolipid, arginine, and proline were identified as the most significantly changed. Specifically, glutamate and arginine metabolism-related metabolites were significantly elevated in U-87 MG LbNOX cells, while they remained relatively low in naive cells (Fig. 2C). Additionally, the pathway impact analysis indicated high rankings for glutamate and glutamine metabolisms in U-87 MG LbNOX cells, as evidenced by their high impact scores and low P-values (Fig. 2D). In the relative quantification of these key metabolites (Fig. 2E), glutamate (~ twofold increase), glutamine (~ threefold increase), threonine (~ twofold increase), and arginine (~ twofold increase) appeared to accumulate upon LbNOX expression. Collectively, these results suggest that LbNOX alters glutamate-related pathways within U-87 MG.

Fig. 2
figure 2

LbNOX alters the intracellular glutamate pathway. An overview of metabolic profiling. (A) Principal component analysis (PCA) of naive and LbNOX-expressing U-87 MG cells. The PCA was performed on 91 metabolites selected from the metabolite profiling, which were significantly altered (± 1.5 < fold-change, P value < 0.05). Pink represents the naive sample group, and green represents the LbNOX expression group (n = 3). (B) Heatmap of the 30 most significantly changed metabolites. The color scale values on the side represent relative intensity. Among the 91 changed metabolites, the top 30 that were most significant and showed the most change were displayed. (C) Graph of pathway impact analysis using 91 altered metabolites. The analysis was performed as described using MetaboAnalyst 5.059. The color represents the ratio of altered metabolites in a given pathway; thus, in pathways with lower p-values, more metabolites were altered. The circle size is proportional to the impact score, ranging from 0 to 1.0. (D) Heatmap of metabolites in the pathways that were most significantly impacted by the pathway analysis. (E) Bar graph of the relative abundance of the metabolite that differs most between samples. Black is the LbNOX sample, and gray is the naive sample. The x-axis represents metabolites and the y-axis represents relative abundance based on mass-spectrometry. Glu, glutamate; Gln, glutamine; Thr, threonine; Asp, aspartate; Arg, arginine.

Full size image

Glutamate release was reduced by LbNOX and mitoLbNOX

Our study identified glutamate as the most abundant amino acid through the analysis of metabolic profiling, and notably, the expression of LbNOX significantly elevated its levels. Given the pivotal role of glutamate in GBM invasion, we are more focused on investigating the impact of glutamate on GBM pathogenesis. We conducted glutamate imaging to assess the release of glutamate from GBM cells. The quantification of glutamate released from individual cells was performed utilizing iGluSnFR, a specialized glutamate sensor36. To stimulate glutamate release from U-87 MG cells, we employed TFLLR, a selective agonist of protease-activated receptor 1 (PAR1), which is known to induce glutamate release in astrocytes through both calcium-dependent and independent mechanisms37,38.

When TFLLR is introduced into a single cell through a pressure-based puffing technique, glutamate is released in a G protein-coupled receptor (GPCR)-dependent manner. The released glutamate subsequently binds to iGluSnFR, leading to the emission of GFP fluorescence (Fig. 3A and B). The mean ΔF/F0 in naive U-87 MG cells was measured at 0.028. However, in LbNOX or mitoLbNOX-expressing U-87 MG cells, there was a significant reduction in glutamate release (Fig. 3C and D).

Fig. 3
figure 3

The release of glutamate from U-87 MG is reduced by LbNOX and mitoLbNOX. (A) Schematic image for glutamate imaging. (B) The representative photographs for glutamate imaging from U-87 MG expressed with iGluSnFR, Glutamate sensor. Scale bar = 50 μm. Fluorescence photographs from U-87 MG before and after TFLLR puffing. (C) Traces of glutamate imaging from naive cells (left, n = 39), cells expressing LbNOX (center, n = 29), and mitoLbNOX (right, n = 18). Scales bars = 0.02 ΔF/F0, 20 s time. The black trace is the glutamate signal for individual cells, and the red one is the average value of the cells. The red triangle is the puffing of 500 μM TFLLR at 20 s. (D. The bar graph shows the ΔF/F0 peak in each condition. An individual dot is data from a single cell. Data are reported as mean with SEM. Data were analyzed by a Tukey’s multiple comparisons test. *P = 0.0408, ***P = 0.0008.

Full size image

Migration, invasion, and proliferation of GBM were reduced by LbNOX or mitoLbNOX

Glutamate is well-established as a critical molecule that facilitates GBM invasion, migration, and proliferation. To investigate the influence of glutamate release, which is modulated by LbNOX and mitoLbNOX, on the pathogenesis of GBM, we conducted a series of experiments, including wound healing, invasion, and proliferation assays15,39. A wound-healing assay was performed to assess the migratory capability of cancer cells (Fig. 4A and B). In comparison to naive U-87 MG cells, there was no discernible difference in the motility of U-87 MG cells expressing LbNOX. The analysis of wound closure after 24 h revealed a recovery rate of 79.69% for naive U-87 MG cells and 74.31% for LbNOX-expressing U-87 MG cells. However, U-87 MG cells expressing mitoLbNOX exhibited a substantial reduction in migratory capacity, with an average wound recovery of 48.97% (Fig. 4A and B). In the transwell invasion assay, designed to evaluate the invasive properties of GBM, LbNOX-expressing U-87 MG cells displayed a 49.38% reduction in invasion compared to naive U-87 MG cells, while mitoLbNOX-expressing U-87 MG cells exhibited an even more pronounced decrease of 68.56% in invasion (Fig. 4C and D). We conducted a proliferation assay to assess the proliferation potential of U-87 MG cells in each experimental group. In comparison to naive U-87 MG cells, the LbNOX-expressing group exhibited a notable 25.31% reduction in proliferation, while the mitoLbNOX-expressing group displayed a somewhat lower reduction of 15.58% (Fig. 4E). This decrease in cell number was consistent with previous findings35. We additionally conducted additional experiments to determine whether these LbNOX or mitoLbNOX-induced effects were reproducible in other GBM cell lines. We used U-373 MG and confirmed protein expression in both groups (Fig. 4F). In the wound healing assay (Fig. 4G and H), invasion assay (Fig. 4I and J), and proliferation assay (Fig. 4K), similar trends were observed as in the previous experiment, but migration was reduced more in the LbNOX expression group than in U-87 MG.

Fig. 4
figure 4

Analysis of motility, invasion, and proliferation in GBM expressing LbNOX and mitoLbNOX. (A) The representative photographs of a wound healing assay from U-87 MG naive cells (left), LbNOX expression cells (center), and mitoLbNOX expression cells (right). These photographs show the assay at 0 h (top) and 24 h (bottom) after the scratch with the cell-free region (yellow line). Scale bar = 500 μm. (B) The bar graph shows the percentage of healing in each condition (naive and LbNOX: n = 15, mitoLbNOX: n = 9). (C) The representative photographs for the invasion assay were obtained by staining U-87 MG with DAPI for each condition. Scale bar = 500 μm. (D) The percentage of invasion was presented as a bar graph in each condition (naive: n = 3, LbNOX and mitoLbNOX: n = 5). (E) The bar graph shows the normalized proliferation O.D values (naive, LbNOX, and mitoLbNOX: n = 16). (F) The representative photographs of LbNOX (top) and mitoLbNOX (bottom) in U-373 MG, with photographs of GFP and DIC. Scale bar = 20 μm. (G–K) Same as (A–E). However, J is displayed as the number of DAPI. All individual dots are data from a single well. All data are reported as mean with SEM. All data were analyzed by one-way ANOVA followed by a Tukey’s multiple comparisons test.

Full size image

Calcium response is reduced by LbNOX or mitoLbNOX in U-87 MG

In addition to glutamate, calcium is recognized as a pivotal factor in glioblastoma invasion6. In light of the observations presented in Fig. 4, we conducted calcium imaging experiments to investigate the impact of LbNOX and mitoLbNOX on the calcium response in U-87 MG cells. Similar to the glutamate imaging experiments, we employed TFLLR as a stimulus to elicit intracellular calcium responses. It is known that PAR1 binds to the Gαq protein, and when it is activated, it can cause a calcium response in the endoplasmic reticulum (ER) of the cell40. For calcium imaging, we utilized Fura-2 AM, a fluorescent dye that is capable of measuring the levels of cytosolic-free calcium41. Figure 5A provides a representative and schematic image illustrating the calcium imaging procedure. On the day of the experiment, Fura-2 AM was introduced into U-87 MG cells in each experimental group. Upon the bath application of TFLLR, there was an observable increase in intracellular calcium levels due to PAR1 activation. The normalized calcium response ratio for naive U-87 MG cells was measured at 0.02. In contrast, in both the LbNOX and mitoLbNOX expression groups, the calcium response was significantly diminished, with values of 0.014 and 0.008, respectively (Fig. 5B and C).

Fig. 5
figure 5

Expression of LbNOX and mitoLbNOX reduces intracellular calcium responses. (A) The representative photographs for calcium imaging from U-87 MG loaded with Fura-2 AM. Photographs of DIC, GFP, and Fura-2 AM are from the left. Scale bar = 50 μm. Schematic image for calcium imaging (center) and pseudo-color photographs from Fura-2 AM loaded U-87 MG before and after TFLLR application (right). The scale bar on the right is the normalized 340/380 nm ratio. (B) Traces of calcium imaging from naive cells (left, n = 115), cells expressing LbNOX (center, n = 105), or mitoLbNOX (right, n = 77). Scale bars = 0.02 ratio, 100 s time. The black trace is the calcium signal for individual cells, and the red one is the average value of the cells. The blue horizontal bar is the application of 30 μM TFLLR by the time of the scale. (C) The bar graph shows the normalized 340/380 nm peak in each condition. An individual dot is data from a single cell. Data are reported as mean with SEM. Data were analyzed by one-way ANOVA followed by a Tukey’s multiple comparisons test. ****P < 0.0001, **P = 0.0026.

Full size image

Rotenone and Nicotinamide Riboside do not affect the invasion of U-87 MG

To further elucidate, we conducted experiments to ascertain whether these effects were associated with the metabolic pathway for ATP production.

Rotenone, a well-known inhibitor of mitochondrial ATP complex I, was employed for this purpose42,43. The ATP assay showed that when rotenone was added to U-87 MG cells, the concentration of ATP within the cells was reduced significantly (measured in mM), going from 1.03 mM in the control group to 0.78 mM in the rotenone treatment group (Fig. 6A). Nicotinamide riboside (NR) is recognized as a precursor to NAD+44. To investigate the effects of NR, we performed NAD+/NADH assays by treating U-87 MG cells with NR and revealed a consistent increase in the NAD+/NADH ratio at all time points when compared to the control (Fig. 6B).

Fig. 6
figure 6

Rotenone and Nicotinamide Riboside do not affect invasion. (A) The bar graph shows the concentration of ATP in each condition (control and Rotenone: n = 6). (B) The NAD+/NADH ratio curve in each condition by time using kinetic mode measurement. Each plot contains 3 wells of data. (C) The representative photographs for the invasion assay were obtained by staining U-87 MG with DAPI for each condition. Scale bar = 500 μm. (D) The percentage of invasion was presented as a bar graph in each condition (control: n = 2, rotenone: n = 7, NR: n = 6). (E) Traces of calcium imaging from cells treated with rotenone (center, n = 108), NR (right, n = 114), and non-treated control (left, n = 78). Scale bars = 0.02 ratio, 100 s time. The black trace is the calcium signal for individual cells, and the red one is the average value of the cells. The blue horizontal bar is the application of 30 μM TFLLR by the time of the scale. The orange one is the application of 1 μM rotenone and the purple one is the application of 500 μM NR during the entire recording. (F) The bar graph shows the normalized 340/380 nm peak in each condition. An individual dot is data from a single well (A, D), and a single cell (F). All data are reported as mean with SEM. Data were analyzed by two-tailed unpaired Student's t test (A, B), and one-way ANOVA followed by a Tukey’s multiple comparisons test (D, F). ****P < 0.0001, ***P < 0.001.

Full size image

Subsequently, we assessed the influence of rotenone and NR on U-87 MG cells by conducting transwell-invasion assays (Fig. 6C and D). Interestingly, there was no observable difference in the invasive capacity between the rotenone and NR-treated groups when compared to the non-treated control (Fig. 6C and D). Further calcium imaging revealed a reduced calcium response in both the rotenone and NR-treated groups (Fig. 6E and F). The normalized calcium response ratio was significantly decreased in both the rotenone-treated and NR-treated groups when compared to the non-treated control group (n = 78) (Fig. 6F). These findings emphasize that while rotenone and NR led to a decreased intracellular calcium response, they did not diminish the invasion ability.

Discussion

To summarize this study, we confirmed that NAD+/NADH in GBM was increased by LbNOX or mitoLbNOX, thereby leading to intracellular glutamate accumulation (Figs. 1 and 2). Glutamate imaging revealed that activity-dependent glutamate release was reduced (Fig. 3). Thus, the migration, invasion, and proliferation of GBM were reduced due to the deficiency of autocrine-acting glutamate (Figs. 4, and 7).

Fig. 7
figure 7

Graphical summary of this study.

Full size image

Through this study, we found that changes in the NAD+/NADH ratio affect the intrinsic functions of GBM cells. The study was investigated under the assumption that NAD+/NADH, as an important metabolic mediator, potentially exerts regulatory influence over metabolically unstable GBM. The effects of the redox reactions involving NAD+ and NADH on intracellular metabolic processes are profound and intricately interconnected. In addition, many researchers have interpreted the role of NAD+ and NADH in the metabolic process, but the profound significance of the NAD+/NADH ratio remains inadequately interpreted, particularly within cancer cells such as GBM. The research conducted by van Horssen et al. in 2013 investigated the biological function of NAD(H) in GBM by modulating NAD(H) levels through the inhibition of NAMPT activity using both shRNA and pharmacological inhibitors30. The absence of NAMPT significantly impacted the alteration in the quantity of NADH or NAD+, thereby influencing the movement of GBM. Nevertheless, there is a lack of evidence about the molecular mechanism of NAD+ and NADH on GBM motility, and it is only determined by their absolute quantities. To address this issue, we implemented precise manipulation of the NAD+/NADH ratio through the utilization of bacterial NADH oxidase and found that the regulation of GBM function is influenced by calcium signaling and glutamate, which are specifically modulated by alterations in the NAD+/NADH ratio.

We discovered that GBM had altered glutamate and glutamine metabolism through metabolite profiling. Glutamate can be typically synthesized through two mechanisms in which the NADH/NAD redox is an essential component of the pathway45. Glutamate is synthesized through a reductive amination of 2-oxoglutarate catalyzed by glutamate synthase (GltS) with glutamine serving as the nitrogen donor where NAD+ is required as a cofactor. An alternative method involves the reductive amination of 2-oxoglutarate using glutamate dehydrogenase (GDH) in conjunction with ammonium as the nitrogen donor and NAD+ is also needed as a hydrogen acceptor46,47. Glutamine metabolism significantly contributes to cancer cell resistance to chemotherapy through several mechanisms. First, glutamine metabolism significantly contributes to cancer cell resistance to chemotherapy through several mechanisms Glutamine is essential for the synthesis of nucleotides, which are critical for DNA and RNA production. This is particularly important in rapidly dividing cancer cells. Inhibition of glutamine metabolism can reduce the availability of nucleotides, thereby affecting the survival of chemotherapy-resistant cells. For instance, cisplatin-resistant cancer cells rely heavily on glutamine for nucleotide biosynthesis, and targeting this pathway can sensitize these cells to chemotherapy48. Second, glutamine metabolism supports the synthesis of glutathione, a major antioxidant that helps cancer cells manage oxidative stress induced by chemotherapy. By maintaining redox homeostasis, glutamine allows cancer cells to survive the oxidative damage caused by chemotherapeutic agents. Enhanced glutamine metabolism has been observed in sorafenib-resistant hepatocellular carcinoma cells, where it supports survival through the NADPH-dependent glutathione redox system49.

We have shown that the degree of healing and invasion ability of GBM is further reduced due to the expression of mitoLbNOX (Fig. 4B and D). This suggests that the changes in the NAD+/NADH ratio within mitochondria caused by mitoLbNOX, rather than at the level of the entire cell, are directly related to GBM invasion. Similarly, there is evidence for the involvement of mitochondria in the cancer cell invasion. Upregulation of the mitochondrial fission protein Drp1 is associated with invasive breast carcinoma and lymph node metastases50. Mitochondrial fragmentation promotes tumor cell invasion in breast cancer51. OXPHOS (Oxidative Phosphorylation) is essential for the invasive and metastatic potential of cancer cells. For instance, overexpression of PGC-1α, a key regulator of mitochondrial biogenesis and OXPHOS, enhances the invasion of breast cancer cells. Conversely, suppression of PGC-1α inhibits their invasive potential and attenuates metastasis52.

We demonstrated that the change in the NAD+/NADH induced by LbNOX or mitoLbNOX leads to an alteration in glutamate secretion (Fig. 3). The roles of NAD+ as an oxidizing agent and NADH as a reducing agent are highly diverse, encompassing cellular constituents and metabolism, including the synthesis of glutamate. Redox reactions including oxidoreductase-mediated hydrogen transfer are particularly dependent on the NAD+ and NADH redox pair. These reactions are necessary for a wide variety of electron-exchange-dependent biochemical processes. In these reactions, the role of NAD+ is that of an electron acceptor, whereas the role of NADH is that of an electron donor. Several processes that require NAD(H) as a coenzyme are related to catabolism and the harvesting of metabolic energy. The oxidized form of NAD, NAD+, serves as a cofactor for enzymes that participate in the following reactions46: (1) alcohol metabolism (alcohol and aldehyde dehydrogenases), (2) glycolysis (glyceraldehyde phosphate dehydrogenase), (3) oxidative decarboxylation of pyruvate acetyl-CoA (pyruvate dehydrogenase), (4) fatty acid β-oxidation (3-hydroxyacyl-CoA dehydrogenase), (5) tricarboxylic acid cycle (α-ketoglutarate, isocitrate and malate dehydrogenases), 6) and the synthesis of glutamate (glutaminase (GLS) and glutamate dehydrogenase).

Using LbNOX and mitoLbNOX, we discovered that increased NAD+/NADH in GBM is associated with decreased glutamate release and aggressive behavior. However, many limitations need to be considered both inside and outside the cell. First, gliomas with IDH (isocitrate dehydrogenase) mutations are characterized by altered α-ketoglutarate metabolism, involving the conversion of glutamate to α-ketoglutarate by glutamate dehydrogenase, which consumes NAD+. Glutamate is derived from glutamine through glutaminase, highlighting its importance in cancer metabolism. Glutamine can affect cancer cell aggressiveness by modulating redox responses and metabolic pathways involving NAD+/NADH and NADP+/NADPH45. The precise role of NAD+/NADH in regulating glutamate secretion is unclear, whether due to changes in glutamate synthesis or modulation of key proteins involved in secretion. Further research is needed to elucidate these mechanisms. Second, multiple investigations have demonstrated that the activity of LDHA, a component of lactate dehydrogenase (LDH), and the release of lactate facilitate the invasion and metastasis of numerous types of cancer cells including glioblastoma53,54,55. LDH produces lactic acid, and the redox reaction of NAD+/NADH mediates this process56. It is necessary to determine whether changes in NAD+/NADH by LbNOX or mitoLbNOX are also involved in this process, and if so, whether they are involved in invasion. In addition, there are many additional issues to be addressed, such as how metabolic pathways are reorganized in GBM in response to glutamate released from neurons surrounding GBM, and how changes in calcium responses via LbNOX and mitoLbNOX affect GBM aggressiveness.

The biological functions of NAD+ have been studied extensively over the past several decades. However, due to the absence of appropriate tools, not many scientific attempts have been made on the biological function of the NAD+/NADH ratio. The importance of changes in the ratio between NADH and NAD+ has recently been highlighted. It has been reported to be involved in insulin resistance, and its biological function differs from the absolute amount of NAD+ and NADH57,58. Furthermore, we found that it inhibits the function of GBM. This highlights the importance of developing anti-diabetic drugs that target the NAD+/NADH ratio, and our research will make further progress in developing medicine for the treatment of GBM.

Methods

Generation of plasmids & lentivirus and infection

The gene sequences of LbNOX and mitoLbNOX were synthesized by Genewiz Corporation. Then, PCR amplification was performed on the pUC vector containing the synthesized gene. The amplified products were cloned into the pHR-CMV-TetO2_3C-Avi-His6__IRES-EmGFP vector (addgene, MA, USA, #113888) using the Gibson Assembly method. To produce lentivirus, pHR-CMV-TetO2_3C-Avi-His6__IRES-EmGFP-LbNOX, and pHR-CMV-TetO2_3C-Avi-His6__IRES-EmGFP-mitoLbNOX were transfected into HEK 293T using Lipofectamine 3000 (Thermo Fisher Scientific, MA, USA, #L3000001) according to the addgene Lentiviral Guide. For glutamate imaging, to prevent fluorescence from overlapping with the glutamate sensor, the insert was cloned into the following vector; pHR-CMV-TetO2_3C-Avi-His6_IRES-mRuby2 (addgene, #113889). Media were collected every 24 h for 3 days post-transfection. After generating lentiviruses, GBM was infected with the lentivirus-containing media. Following this, culture media were exchanged after 24 h to help cell growth and virus expression efficiency.

Cell culture

The human glioblastoma cell line [U-87 MG] was purchased from ATCC (VA, USA, #HTB-14), and [U-373 MG] was purchased from Korean Cell Line Bank (Seoul, Korea, #30017). Cells were maintained in DMEM (CORNING, NY, USA, #10-013-CVRC) supplemented with 10% FBS (gibco, NY, USA, #16000-044), and 1% Penicillin–Streptomycin Solution (10,000 units/mL Penicillin, 10,000 µg/mL Streptomycin) (cytiva, Seoul, Korea, #SV30010). All cells were grown at 37 °C in a humidified atmosphere containing 5% CO2.

NAD+/NADH assay

Intracellular NAD+/NADH ratios were measured using a NAD/NADH Assay KIT II (Colorimetric) (Abcam, Cambridge, England, #ab221821). 1.2 × 106 U-87 MG were seeded on culture plates in each condition, and the assay was performed after 24 h of incubation. 500 µM nicotinamide riboside (NR) (Sigma-Aldrich, MO, USA, #SMB00907) was treated for 24 h. The cells were harvested using 0.05% Trypsin–EDTA (gibco, NY, USA, #25300-054). Assay and ratio calculations were performed according to the manufacturer’s instructions. OD450 nm was measured in kinetic mode at 10-min intervals for 90 min using a spectrophotometer (BMG LABTECH, Ortenberg, Germany, #SPECTROstar Nano). Data were obtained at times when the standard curve was significant.

ATP assay

Intracellular ATP concentrations were measured using an ATP assay kit (Abcam, Cambridge, England, #ab83355). 1.2 × 106 U-87 MG were seeded on culture plates in each condition, and the assay was performed after 24 h of incubation. 1 µM rotenone (Sigma-Aldrich, MO, USA, #R8875) was treated 6 h after seeding and incubated for 24 h. The cells were harvested using 0.05% Trypsin–EDTA (gibco, NY, USA, #25300-054). Colorimetric ATP assay and concentration calculations were performed according to the manufacturer's instructions. OD 570 nm was measured using a spectrophotometer (BMG LABTECH, Ortenberg, Germany, #SPECTROstar Nano).

Metabolic profiling

LC–MS analysis for metabolic profiling was conducted using Agilent time-of-flight (TOF) mass spectrometry coupled with the Agilent 1290 liquid chromatography system. Extracted metabolites were separated using a Cogent diamond hydride type C column. The gradient elution of metabolites was conducted using solvent A (DDW with 0.2% formic acid) and solvent B (acetonitrile with 0.2% formic acid). Detailed conditions for gradient elution are as follows: 0–2 min, 85% B; 3–5 min, 80% B; 6–7 min, 75%; 8–9 min, 70% B; 10–11.1 min, 50% B; 11.1–14 min 20% B; 14.1–24 min 5% B followed by a 10 min re-equilibration time at a flow rate of 0.4 mL/min. Concentrations of detected metabolites were normalized by the protein concentrations of the respective samples measured by the Pierce BCA protein assay kit. Metabolites were identified using Agilent Profinder B.06.00 software. Metabolites with fold change > 1.5 and P value < 0.05 were selected for principal component analysis (PCA) and pathway analysis. PCA and pathway analysis were conducted using the online open-source software Metaboanalyst 5.0. In the pathway analysis, P values were calculated based on the number of altered metabolites in a given pathway, where a higher ratio of altered metabolites generates a lower p-value. Meanwhile, the impact score is calculated based on the number of connections a metabolite has as a metabolic node within a given pathway59.

Wound healing assay

Cells were grown to full density in culture plates. The monolayer was scratched using a 10 µL pipette tip and each location at 0 h was marked and photographed. After 24 h of incubation, photographs were taken of the same location. Photographs were taken at 100× magnification (KOREA LAB TECH, Seongnam, Korea, #KCS3-160S, #KI-400), and cell-free areas were measured using Image J (Ver. 1.53k). Data are expressed as a percentage of the filled area after 24 h in one well compared to 0 h.

Invasion assay

1.0 × 105 U-87 MG for each condition were seeded on transwell plates (SPL, Pocheon, Korea, #35224) with vitrogel hydrogel matrix (The Well Bioscience, NJ, USA, #VHM01). 2% serum media was put inside the transwell, and 20% serum media was placed outside the transwell. Cells were incubated for 48 h. 1 µM rotenone (Sigma-Aldrich, MO, USA, #R8875) & 500 µM nicotinamide riboside (NR) (Sigma-Aldrich, MO, USA, #SMB00907) were treated for 48 h. DPBS (WELGENE, Gyeongsan, Korea, #LB001-01), 3.7% formaldehyde (Sigma-Aldrich, MO, USA, #252549), methanol (Samchun Chemicals, Seoul, Korea, #M0588), and DAPI (Thermo Scientific, MA, USA, #62248) were used for washing, fixation, permeabilization, and staining, respectively. Fluorescence photographs were obtained with the following microscope and camera at 100× magnification (Olympus, Tokyo, Japan, #BX51, #U-HGLGPS, #U-LH100HG). Invaded areas were measured using Image J (Ver. 1.53k). Three photographs were randomly taken per well, and the average of the three regions was compared with the average of naive cell wells, and the data were expressed as the percentage of invasion. In additional U-373 MG experiments, cell counts were performed for more accurate quantification.

Proliferation assay

Cell proliferation was measured using a Quanti-Max WST-8 Cell Viability Assay Kit (BIOMAX, Guri, Korea, #QM1000). 1 × 105 U-87 MG were seeded on culture plates in each condition. After 24 h of incubation, the proliferation assay was performed according to the manufacturer’s instructions. OD 450 nm was measured using a spectrophotometer (BMG LABTECH, Ortenberg, Germany, #SPECTROstar Nano). Normalized OD values were obtained based on naive by randomly pairing naive, LbNOX, and mitoLbNOX into one set.

Calcium & glutamate imaging

For imaging, all cells were cultured at similar densities on poly-D-lysine (Sigma-Aldrich, MO, USA, #P6403)-coated 12 mm Ø cover glasses (Paul Marienfeld, Lauda-Königshofen, Germany #0111520). 1 µM rotenone and 500 µM nicotinamide riboside (NR) treatment groups were incubated overnight. For calcium imaging, all cells were incubated in an external solution containing 5 μM Fura-2 AM (Invitrogen, MA, USA, #F1201) plus 1 μM pluronic acid (Sigma-Aldrich, MA, USA, #P2443) for 40 min at room temperature protected from light. For glutamate imaging, iGluSnFR, a glutamate sensor, was expressed in cells via Effectene (Qiagen, Hilden, Germany, #301425) 24 h after lentivirus infection. TFLLR puffing was performed at 10 psi and the size of the glass pipette used for TFLLR puffing is 4–9 mΩ. The external solution contained 150 mM NaCl (Sigma-Aldrich, MO, USA, #S9888), 10 mM HEPES (Carl Roth, Karlsruhe, Germany, #230-907-9), 3.2 mM KCl (Sigma-Aldrich, MO, USA, #P9541), 2 mM CaCl2 (Sigma-Aldrich, MO, USA, #C3881), 2 mM MgCl2 (Sigma-Aldrich, MO, USA, #M2670), 22 mM sucrose (Carl Roth, Karlsruhe, Germany, #200-334-9), and 5.6 mM glucose (Sigma-Aldrich, MO, USA, #G8270), and pH was adjusted to 7.4. Intensity photographs were obtained with the following microscope and camera at 200 × magnification (Nikon, Tokyo, Japan, #TE300) (HAMAMATSU, Shizuoka, Japan, #C11440-36U) (CoolLED, Andover, England, #pE-340fura). Filters used for glutamate and calcium imaging were as follows; 49054 ET-GFP for 455–465 nm LEDs (CHROMA, USA, #49054), and 79001 ET-Fura 2 (CHROMA, USA, #79001). The pE-340fura (light source) power was used at a maximum of 36 W, with an output of 340 nm and 380 nm for calcium imaging, and 450–650 nm for glutamate imaging. Calcium imaging was conducted with an integration time of 0.01 s, and glutamate imaging with an integration time of 0.05 s. Imaging Workbench version 9.0.14.0 was used for the acquisition of intensity photographs and ratio calculations. TFLLR (synthesized by PEPTRON, Daejeon, Korea), NR (Sigma-Aldrich, MO, USA, #SMB00907), and rotenone (Sigma-Aldrich, MO, USA, #R8875), were used for imaging. To normalize the individual fluorescence intensity of cells, ΔF/F0 or normalized 340/380 ratio was calculated as follows. Each value was recorded at 1-s intervals from the Imaging Workbench. ΔF/F0 or 340/380 ratio was calculated using the following formula: \(\frac{(V-{AVR}_{Base})}{{AVR}_{Base}}\). \(V\): value for each time, \({AVR}_{Base}\): Average of Value before applicating TFLLR.

Statistical data analyses

In the results of Glutamate and Calcium imaging, all ‘n’ were obtained from single cells. The ‘n’ obtained from NAD+/NADH assay, ATP assay, Wound healing assay, Invasion assay, and Proliferation assay is the result from a single well. Image J (Ver. 1.53k) was used for photograph area analysis. All bar graphs, trace figures, and statistical analyses were obtained using GraphPad Prism (Ver. 9.4.1). When there were two comparison subjects, a two-tailed unpaired Student's t-test was performed, and when there were more than two, one-way ANOVA followed by a Tukey’s multiple comparisons test was performed. All data are reported as mean with SEM. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, and ns means not significant.