Bioenergetic modulators hamper cancer cell viability and enhance response to chemotherapy

Abstract Gliomas are characterized by a marked glycolytic metabolism with a consequent production of massive amounts of lactate, even in the presence of normal levels of oxygen, associated to increased invasion capacity and to higher resistance to conventional treatment. This work aimed to understand how the metabolic modulation can influence tumour aggressive features and its potential to be used as complementary therapy. We assessed the effect of bioenergetic modulators (BMs) targeting different metabolic pathways in glioma cell characteristics. The in vivo effect of BMs was evaluated using the chicken chorioallantoic membrane model. Additionally, the effect of pre‐treatment with BMs in the response to the antitumour drug temozolomide (TMZ) was analysed in vitro. Cell treatment with the BMs induced a decrease in cell viability and in migratory/invasion abilities, as well as modifications in metabolic parameters (glucose, lactate and ATP) and increased the cytotoxicity of the conventional drug TMZ. Furthermore, all BMs decreased the tumour growth and the number of blood vessels in an in vivo model. Our results demonstrate that metabolic modulation has the potential to be used as therapy to decrease the aggressiveness of the tumours or to be combined with conventional drugs used in glioma treatment.

present limited efficacy, leading to tumour relapse and poor patient survival rates. 5 Temozolomide (TMZ) is a first-line oral alkylating drug used in glioma treatment, being its cytotoxicity based on TMZgenerated O6-methylguanine-DNA adducts. However, the DNA damage induced by TMZ can be repaired by the O6-methylguanine-DNA methyltransferase (MGMT) repair enzyme, which is associated with TMZ therapy resistance and treatment failure. 9,10 Therefore, it is important to develop more specific and effective therapies targeting glioma features, such as the reprogrammed metabolism. 11 The glycolytic enzymes, specifically overexpressed in cancer cells, are one of the main targets in this field and several compounds targeting glycolysis are already in clinical trials. 12 Dichloroacetate (DCA) is a pyruvate dehydrogenase kinase (PDK) inhibitor that redirects cell metabolism towards OXPHOS. PDK is a direct inhibitor of pyruvate dehydrogenase (PDH), a key enzyme that shifts the flux of pyruvate into mitochondria to promote OXPHOS. Many reports showed the promising effect of DCA in cancer therapy in in vitro and in vivo cancer models, [13][14][15] although aspects such as its toxicity and dose limit effects are still unclear. 16,17 Other glycolytic inhibitor with potential anticancer activity is 2-deoxy-D-glucose (2-DG). 2-DG is a glucose analogue that competes with glucose in the first step of glycolysis, being converted to deoxyglucose-6-phosphate, a molecule that cannot be further metabolized, inhibiting hexokinase 2 (HK2), thus blocking glycolysis and the pentose phosphate pathway. 18 2-DG is described as being able to induce tumour cell death in different type of cancers. [18][19][20][21] Although the potential use of glycolytic inhibitors in cancer therapy, recent studies have demonstrated that in brain tumours, mitochondrial oxidation is also an important pathway in metabolism to support the rapid cell growth. 22 Some studies have demonstrated that biguanides, used commonly in diabetes treatment and that act on OXPHOS, may also have antitumour action. Phenformin is an analogue of metformin that exhibits a larger antitumour activity in lung, 23 breast 24 and colorectal cancers. 25 Recently, it has been described that the compounds that target the mitochondria can also affect glycolysis and vice versa. For instance, metformin, which inhibits the complex I of the mitochondria respiratory chain, can also target HK2. 26 Therefore, the aim of this study was to understand the importance of metabolic inhibition in glioma proliferation and aggressiveness, and how bioenergetic modulators (BMs), such as DCA, 2-DG and phenformin, can be potentially used as antitumour drugs, namely as combined therapy. There are very few reports describing the metabolic behaviour of glioma cells under the conditions of this study, as well as the use of these metabolic modulators as coadjuvants of the standard treatment, TMZ. U251 and SW1088, two high-grade glioma cell lines, were obtained from American Type Culture Collection. The immortalized astrocyte cell line hTERT/E6/E7 HOXA9, previously retrovirally infected with MSCVneo vectors containing HOXA9 cDNA, 27 was kindly provided by Professor Bruno Costa, ICVS, University of Minho. All cell lines were maintained in Dulbecco's Modified Eagle's Medium (DMEM) supplemented with 10% foetal bovine serum (FBS) and 1% penicillinstreptomycin solution, at 37°C and 5% CO 2 .

| Drugs
TMZ (Sigma-Aldrich) was dissolved in dimethyl sulfoxide (DMSO) in a 100 mmol/L stock solution, from which the working solutions were prepared. The BMs 2-DG, DCA and phenformin (Sigma-Aldrich) were dissolved in PBS, to prepare stock solutions of 1000, 10 000 and 100 mmol/L, respectively, from which the working solutions were prepared.

| Cell viability assay
Cells were plated into 96-well plates, at a density of 3 9 10 3 cells/ well for all cell lines, for TMZ, 2-DG and phenformin exposure, during 72 hours and DCA exposure, during 48 hours. After treatment, cell viability was determined by the sulforhodamine B (SRB) assay, as described previously. 28 IC 50 values were estimated from at least 3 independent experiments, each one in triplicate, using the GraphPad Software.

| Colony-forming assay
Five hundred cells were seeded in 6-well plates and treated with the IC 50 of the different BMs, during the respective incubation times.
Untreated cells were used as control. After incubation, the medium containing the compounds was removed, cells were washed twice with PBS and then fresh medium was added. Cells were then allowed to grow for 10 days. The formed colonies were fixed for 5 minutes with 3.7% (w/v) paraformaldehyde in PBS and stained for 20 minutes with 0.05% (w/v) violet crystal in distilled water. The plating efficiency (PE) was calculated as the percentage of the number of grown colonies over the number of cells seeded in the control before BMs treatment. For each condition, the survival fraction was determined as the number of colonies over the number of cells seeded 9 1/PE.

| Metabolism assays (Extracellular glucose and lactate, and ATP content)
Cells were plated in 6-well plates at a density of 3 9 10 5 cells/well. Then, cells were treated with the respective BMs IC 50 , and the cell culture medium was collected after the respective incubation time, for glucose and lactate quantification. Glucose and lactate were quantified using commercial kits (SPINREACT), according to the manufacturer's protocols and normalized against total biomass. Untreated cells, incubated during the same period of time, were used as control. Results are expressed as total lg of metabolite/total biomass. Simultaneously, the cells were used for protein extraction and to TAVARES-VALENTE ET AL. | 3783 quantify the intracellular ATP using a commercial kit (Molecular Probes), according to the manufacturer's instructions. The ATP content was normalized for the concentration of protein and also against the value obtained with untreated cells, incubated during the same period of time of the respective treatment, set as 1. The results presented correspond to the average of at least three independent experiments.

| Wound-healing assay
Cells were plated in 6-well plates at a density of 1 9 10 6 cells/well and the wound-healing assay was performed. Cells were treated with different concentrations of BMs for 24 hours, and the wound areas were photographed at 0, 12 and 24 hours. The relative migration distances were analysed using Image J Software. The relative migration for glioma cells treated with the respective BM IC 50 was compared with the control (untreated cells).

| Invasion assay
Cell invasion in cancer cell lines was analysed using 24-well BD Bio-

| Immunohistochemistry
Immunohistochemistry for Ki67 (AP10243CM, Gennova) was performed according to the avidin-biotin-peroxidase principle (R.T.U. Vectastin Elite ABC kit; Vector Laboratories), as previously described by our group. 6 In brief, deparaffinized and rehydrated slides were submitted to heat-induced antigen retrieval in the microwave for 15 minutes with 10 mmol/L citrate buffer (pH 6.0). After endogenous peroxidase inactivation, incubation with the primary antibody was performed for 2 hours at room temperature. The immune reaction was visualized with 3,3 0 -Diamonobenzidine (DAB + Substrate System; Dako). All sections were counterstained with Gill-2 haematoxylin. For negative controls, primary antibodies were replaced by a universal negative control antibody (N1699, Dako).

| Chicken chorioallantoic membrane (CAM) assay
In brief, fertilized chicken eggs were incubated at 37°C. On day 3 of development, a window was made into the eggshell after puncturing the air chamber, and eggs were sealed with BTK tape and returned to the incubator. On day 9 of development, the

TMZ cytotoxicity
1.5 9 10 3 cells/well were seeded into 96-well plates and pretreated with a fixed concentration of BMs (a concentration previously determined that increase cytotoxicity of TMZ but do not induce cell death per se), during the respective incubation times.
In untreated cells, the medium was replaced at this time-point.
After the period of incubation, the medium containing the compounds was removed, cells were washed twice with PBS and the cells pre-treated and not pre-treated with BMs were exposed to TMZ at the same range of concentrations previously used for 72 hours. The effect of TMZ alone and BM + TMZ on cell viability was evaluated using the SRB assay. Additionally, the action of BMs on TMZ effect on cell migration and colony formation ability was evaluated through wound-healing and colony formation assays, respectively. In both assays, cells were pre-treated or not with a fixed concentration of BMs (2-DG (5 mmol/L), DCA (20 mmol/L) and phenformin (0.01 mmol/L)-concentrations previously determined that increased cytotoxicity of TMZ but did not induce cell death per se) for the respective time of incubation. After this period of incubation, cells were treated with 100 lmol/L of TMZ for 72 hours in the colony formation assay and until 48 hours in the wound-healing assay. The assays continued as described previously in this materials and methods section.

| Statistical analysis
The GraphPad prism 5 software was used, with the Student's t-test, considering significant values to be P ≤ .05.

| Treatment with bioenergetic modulators affects cell survival and changes the metabolic profile of glioma cells
Cell behaviour of two glioma cell lines (U251 and SW1088) was evaluated on metabolic remodelling using different metabolic modulators, 2-DG, DCA and phenformin. A decrease in cell viability was triggered in both cell lines by the three compounds, in a dose-and time-dependent manner ( Figure 1A). The lower IC 50 values were found for phenformin (<1 mmol/L for both cell lines), whereas 2-DG IC 50 values were in the range of 10-35 mmol/L and DCA IC 50 values were higher than 100 mmol/L ( Figure 1B). Accordingly, all the compounds induced a decrease in the ability of both cell lines to form colonies, after a recovery period in the absence of compounds ( Figure 2A). In U251 cells treated with 2-DG and DCA, not only the colony number was lower but also the colony size. In contrast, the phenformin effect in the decreased ability to form colonies was more noticeable in SW1088, although the effect in the size was not so evident ( Figure 2A, B). As referred previously, tumour cells present a metabolic reprogramming, being most of cellular ATP generated from glucose via aerobic glycolysis ("Warburg effect") rather than by OXPHOS, what leads to a high rates of lactic acid production. However, most tumours do not depend completely of glycolysis for ATP supply, as mitochondrial metabolism is not decreased in all cancer cells. 29 To analyse the metabolic profile in glioma cells, intracellular ATP content and extracellular lactate and glucose levels were measured for both cell lines treated with the respective BM IC 50 values. Untreated cells, incubated for the same period of time, were used as control. It was observed that the glycolytic inhibitors, 2-DG and DCA were able to reduce the consumption of glucose, as well as the production of lactate and ATP content, in both cell lines, although more evident for U251 cells, especially for DCA ( Figure 3). Different results were obtained when the glioma cells were treated with phenformin. In U251 cell line, a significant increase in glucose and lactate extracellular levels was observed, as well as ATP production, in opposition to what happened with the other BMs. This effect can be due to an activation of energy production through glycolytic pathway when OXPHOS was inhibited. In SW1088 cells, glucose consumption was not altered and lactate production was higher compared to the control. Concerning ATP content, the blockage of this pathway impaired the energy production by this cell line. This indicates that SW1088 present a markedly oxidative phenotype, as inhibition by phenformin induced the most relevant alterations in the metabolic profile of this cell line. In contrast, U251 cells present a more glycolytic phenotype, as the glycolytic inhibitors induced a higher decrease in glucose consumption, and lactate/ATP production.

| Metabolic modulation reduces the migration and invasion capacity of glioma cells
To understand the influence of metabolic modulation on the motility of glioma cell lines, we assessed cell migration by the wound-healing assay, 12 and 24 hours after the treatment with BMs. Concerning U251 cells, all metabolic inhibitors induced a decrease in the motility, compared to untreated cells ( Figure 4A). Nevertheless, phenformin was the less effective in this reduction. In SW1088 cells, metabolic modulation was not able to reduce the migratory ability in a significant way, as it presented already a low ability to close of wound, after 12 hours. However, after 24 hours, 2-DG and phenformin induced a reduction in cell migration capacity. It was also observed that the invasion ability was affected by metabolic modulation ( Figure 4B). This inhibition was more evident in U251 cells, with the glycolytic modulators, as they present a higher basal invasion capacity, compared to SW1088 cells. Concerning phenformin, it was not able to induce a significant decrease in invasion. Concerning SW1088 cells, phenformin induced a higher reduction in the invasive capacity, compared with glycolytic inhibitors, namely DCA.

| Treatment of cancer cells with the metabolic modulators decreases glioblastoma proliferation in vivo
According to our in vitro results, U251 cells exhibit higher glycolytic rates, compared with SW1088 cells. However, all the compounds including phenformin were able to decrease cell migration/invasion and cell proliferation of these cells. Therefore, we aimed to evaluate

| Metabolic inhibition potentiates temozolomide cytotoxicity
This study also aimed to investigate the influence of BMs, acting at different metabolic targets, on the efficacy of the conventional antitumour drug TMZ, with the objective to overcome the treatment resistance commonly developed during therapeutic regimens. As described previously, the main mechanism of resistance in glioma is increased MGMT activity. Both glioma cell lines used did not express TAVARES-VALENTE ET AL.  Figure S4).

| DISCUSSION
Although the knowledge of the influence of the bioenergetic status on tumour characteristics increased greatly in the last years, there is a relatively modest knowledge on the efficacy of metabolic inhibitor compounds, namely in the clinical setting. One of the most aggressive and lethal types of brain human cancer is glioblastoma.
However, the efficacy of the current therapies is very modest due to the development of multidrug resistance (MDR) phenotype, together with the disease recurrence. 30 The switch of metabolism present in gliomas, with an increase in glycolysis as main energy source, is correlated with a worse prognostic and failure of antitumour therapies. 7 In this study, we intend to understand the role of the reprogrammed metabolism and how its modulation can improve the conventional existent therapies. Many reports showed that the use of metabolic inhibitors potentiate the antitumour therapy and reduce tumour aggressiveness [35][36][37] Despite the existence of many Regarding SW1088 cells, our results indicate that both glycolysis and OXPHOS contribute to energy production. Indeed, all the agents led to a decrease in the ATP content in a similar way.
About the glycolytic inhibitors, the effect was greater when 2-DG was used, what is particularly evident in the glucose levels in the culture medium. Concerning phenformin, SW1088 cells were able to rely on glycolysis when the OXPHOS is inhibited, as glucose consumption was maintained in levels similar to the control and lactate production was higher. However, ATP production was lower, revealing that glycolysis is less effective as energy source. Moreover, we found that metabolic inhibition can also decrease the migration/invasion abilities of tumour cells, important hallmarks in the first steps of the metastatic process. We can hypothesize that the metabolic inhibition reduced lactate production and consequently decreased lactate and proton efflux, increasing extracellular pH (pHe), and motility is compromised. Our research group showed that blockage of lactate efflux, through MCT inhibition decreased the migration and invasion abilities in different cancer models. 6,49 Additionally, we observed that metabolic inhibition, by the use of glycolytic inhibitors, induced an increase in pHe compared to untreated cells (Table S1). Nevertheless, our results also demonstrated that when the cells were exposed to phenformin, the pro- Regarding all the results obtained, it is important to note that we demonstrated that BMs decreased cell proliferation and survival, even when used alone, having the potential to be used as an alternative therapy in glioma, as they are able to cross the brain-blood barrier (BBB). 29 F I G U R E 8 Effect of 2-DG (5 mmol/L), DCA (20 mmol/L) and phenformin (Phen. 0.01 mmol/L), pre-treatment on TMZ effect on cell migration and cell colony formation. All the cell lines, U251, SW1088 and hTERT/E6/E7 HOXA9 cells were pre-treated or not with fixed concentrations of BMs during the respective incubation time, followed by the incubation with 100 lmol/L of TMZ. A, Cell migration was quantified by the wound-healing assay, after pre-treatment with the BMs and during incubation with TMZ. Pictures were taken at 409 magnification (migration) and 2009 magnification (invasion) in a Nikon eclipse TE 2000-U microscope. B, Representative pictures of colony formation in both cell lines. Cells were incubated with different compounds, and after a period of recovery for 10 days without compounds, the survival fraction was calculated. Pictures were taken at 2009 magnification in a Nikon eclipse TE 2000-U microscope. Results represent the mean + SEM of at three independent experiments. *P < .05; **P < .01; ***P < .001 compared to cells treated with TMZ alone. # P < .05; compared to untreated cells (control) Ns, no significant imidazole-4-carboxamide (MTIC) seems to be pH dependent (pH > 7). However, there is some controversy on their mechanism of action and additionally, the reports did not specify the place where this activation occurs, inside or outside of tumour cells. 53 Deprivation of tumour energy may predictably potentiate conventional chemotherapeutic treatments, as many of the proteins associated with the MDR phenotype are energy dependent, such as the ATP-binding cassette (ABC) transporter family responsible for the efflux of different high affinity substrates, namely a wide range of antitumour drugs. 57 In fact, some reports showed the overexpression of these resistance proteins either in BBB or in glioma cells, preventing TMZ cytotoxicity. [58][59][60] By Western blot, we demonstrated a strong expression of Pgp in glioma cell lines used in this study (U251 and SW1088) compared to the hTERT/ E6/E7 HOXA9 cells ( Figure S5). However, this is not the case for U251 cells treated with phenformin, where we observed an increase in ATP intracellular content. Nevertheless, phenformin was also able to alter cancer characteristics and to potentiate the cytotoxicity of TMZ, what can be probably explained by other mechanisms. In fact, the MDR phenotype involves different and complex mechanisms that can be used as target to overcome the low efficacy of employed treatment regimens, and metabolic inhibition could be one of the strategies that can be used for this purpose.
Collectively, our findings provided new insights into cancer cell metabolism as a promising therapeutic strategy for patients with gliomas, increasing therapeutic sensitivity. The use of different metabolic inhibitors combined to standard therapy used in clinical routine, reducing therapeutic doses and consequently decreasing adverse effects in normal brain, could be a new helpful option.

CONFLI CT OF INTEREST
The authors confirm that there are no conflicts of interests.

AUTHOR CONTRI BUTIONS
DT-V participated in the conception, design and writing of the manuscript, acquisition, analysis and interpretation of data, as well as development of methodology; SG performed some of the in vivo assays; OQ and FB participated in the conception of the study, data interpretation, and revision of the manuscript and work supervision.