1. IntroductionIn addition to the accelerated glycolytic flux described by Warburg almost a century ago [1,2,3,4,5], a cancer cell utilizes large amounts of glutamine to balance the high need for energy and building blocks asked by its reductive biosynthesis despite harsh microenvironmental conditions [6]. The high proliferation rate, indeed, implies a large use of glutamine g-nitrogen to synthesize nucleotides and non-essential amino acids [7,8,9]. Likewise, cancer growth also implies the high-rate synthesis of fatty acids and thus a high-rate of acetyl-CoA generation warranted by the citrate export from the tricarboxylic acid cycle (TCA) to the cytosol [10]. Carboxylation of pyruvate to oxaloacetate is inhibited in the mitochondria of many cancers, raising the need to compensate for the high efflux of carbon equivalents by enhancing the influx of glutamine-derived glutamate and its conversion to alpha-ketoglutarate [11].In addition to its interference with mitochondrial function, glutamine can also regulate glucose metabolism in the cytosol [12]. Indeed, its processing by glutamine-fructose-6-phosphate transaminase fuels the generation of uridine-diphosphate-N-acetyl glucosamine (UDP-N-Ac-Gla) that diverges glucose flux to the pentose phosphate pathway [PPP] by inhibiting the catalytic function of the glycolysis rate-limiting enzyme phospho-fructo-kinase [12]. This pathway would thus configure glutamine as a primary regulator of PPP activity in cancer cells. In agreement with this observation, amino acid withdrawal has been shown to induce evident redox damage and to decrease the proliferation rate of cancer cells despite their increased expression of the PPP-triggering enzyme glucose-6P-dehydrogenase (G6PD) [13].In addition to its role in fueling the ribose-5-phosphate for nucleic acid synthesis, PPP represents the main provider of NADPH-reductive power for antioxidant responses and the synthesis of fatty acids needed by the membrane generation of a rapidly growing tumor mass. In the present study, we thus verified whether, through which pathways, and to what degree, glutamine regulates glucose-6-phophate (G6P) flux through the PPP. For this purpose, we extended our analysis beyond the well-recognized G6PD-triggered metabolism to consider its largely autonomous and apparently independent counterpart, triggered by hexose-6-phosphate dehydrogenase (H6PD) and confined within the endoplasmic reticulum (ER) [14,15]. Indeed, H6PD expression and catalytic function have been found to play a mandatory for the high proliferation rate and migratory potential of breast cancer cells [16], in agreement with its ER confinement that configures this enzyme as a fundamental regulator of ER–mitochondria networking [17,18] and Ca2+ exchanges [16].

Evaluating the relative contribution of cytosolic and ER-PPPs, here we show that the decelerated proliferation rate induced by glutamine shortage is associated with selective impairment of ER-PPP, eventually resulting in mitochondrial dysfunction due to significant alteration of the ER–mitochondria contact points.

2. Materials and Methods 2.1. Cell Culture

MCF7 and MDA-MB-231 (MDA) cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM, Gibco, Los Angeles, CA, USA) supplemented with 10% fetal bovine serum, 25 mM glucose, 2 mM glutamine, 1 mM pyruvate, 1% penicillin-streptomycin (Gibco, Grand Island, NY, USA) at 37 °C with 5% CO2.

For each experiment, cells were seeded using a control DMEM medium. Five hours after cell seeding, when cells became adherent, the cell cultures were exposed to either fresh control medium (CTR) or DMEM containing only 0.5 mM glutamine (plus all other above supplementations) (LG). Cells were harvested for measurements 48 h later.

2.2. Cell Viability and Proliferation Assays

Cell growth of the tumor cell line cultures was assessed by counting trypan blue-negative cells.

Cell viability was evaluated by propidium iodide (PI) exclusion assays. Cells were stained with 1 μg/mL PI (Enzo Life Sciences, Executive Blvd, Farmingdale, NY 11735, USA) and PI fluorescence was measured using a FACScan Flow Cytometer (Becton Dickinson, Milan, Italy).

Cell proliferation rate was evaluated using carboxyfluorescein diacetate succinimidyl ester (CFSE) (by Sigma incorporated in Merck KGaA, Frankfurter Str. 250, 64293 Darmstadt, Germany). Briefly, cells were plated in 6-well plates and 24 h later adherent cells were labeled in prewarmed phosphate-buffered saline (PBS) containing CFSE at a concentration of 0.3 μM. Cells were incubated for 20 min at 37 °C. After replacing the labeling solution with a fresh prewarmed cell culture medium, cells were incubated for another 20 min at 37 °C to remove unincorporated CFSE. The cells were then cultured either with fresh standard DMEM medium (CTR) or low medium LG. After 24 and 48 h, cells were detached from the substrate by trypsinization, and CFSE fluorescence was acquired on a FACSCan flow cytometer. Flow cytometry data were analyzed by FlowJo software (Version 8.7, Becton–Dickinson, Franklin Lakes, NJ, USA).

2.3. Seahorse Analysis

MDA and MCF7 oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) were determined using a Seahorse XFp Extracellular Flux Analyzer (Agilent Technologies, Santa Clara, CA, USA). A quantity of 4000 or 7000 cells/well (for MDA or MCF7, respectively) were seeded in XFp plates and incubated in a control medium. After 5 h, the medium was replaced, and cells were incubated with either control or LG medium. After 48 h, OCR and ECAR were monitored according to the manufacturer’s instructions. Briefly, the incubation medium was replaced with Agilent Seahorse DMEM, pH 7.4, enriched with nutrients mimicking the respective incubation medium conditions. Three measurements of OCR and ECAR were taken for the baseline and after sequential injection of the selective glutaminase inhibitor Bis-2-(5-phenylacetamido-1,3,4-thiadiazol-2-yl)ethyl sulfide (BPTES, 3 µM), the ATP-synthase inhibitor Oligomycin A (1.5 µM), the ATP synthesis uncoupler carbonyl cyanide-4-trifluoromethoxyphenylhydrazone (FCCP, 1.25 µM), and the Complex I + Complex III inhibitors (rotenone + antimycin A, 0.5 µM).

OCR was expressed as picomol O2 × min−1/million cells of cells while ECAR was converted to proton efflux rate (PER) using WAVE software (Version 2.4, Agilent) after evaluating the buffer factor of the control and LG medium. To calculate the nanomoles of glucose used through glycolysis, PER value expressed in picomol H+ × min−1/million cells of cells was converted to glucose nanomoles × min−1/million cells, applying the equation Glucose + 2 ADP + 2 Pi --> 2 Lactate + 2 ATP + 2 H2O + 2 H+.

2.4. Metabolites Extraction

Cells, seeded in 6-well plates, were quickly rinsed with NaCl 0.9% and quenched with 500 μL ice-cold 70:30 acetonitrile:water. Plates were placed at −80 °C for 10 min, then cells were collected by scraping and sonicated 5 s for 5 pulses at 70% power twice. Samples were centrifuged at 12,000× g for 10 min and supernatants were collected in a glass insert and dried in a centrifugal vacuum concentrator (Concentrator plus/Vacufuge plus, Eppendorf) at 30 °C for about 2.5 h. Samples were then resuspended with 150 μL H2O before analyses.

2.5. Liquid Chromatography with Tandem Mass Spectrometry Metabolic Profiling

Liquid chromatography with tandem mass spectrometry (LC-MS) was performed using an Agilent 1290 Infinity ultrahigh performance LC (UHPLC) system and an InfintyLab Poroshell 120 PFP column (2.1  ×  100 mm, 2.7 μm; Agilent Technologies) coupled with a quadrupole time-of-flight hybrid mass spectrometer (Agilent 6550 iFunnel Q-TOF) and equipped with an electrospray Dual JetStream source operated in negative mode.

The injection volume was 15 μL, and the flow rate was 0.2 mL/min with column temperature set at 35 °C. Both mobile phases A (100% water) and B (100% acetonitrile) contained 0.1% formic acid, the injection volume was 15 μL, and LC gradient conditions were: 0 min: 100% A; 2 min: 100% A; 4 min: 99% A; 10 min: 98% A;11 min: 70% A; 15 min: 70% A; 16 min: 100% A with 5 min of post-run. The flow rate was 0.2 mL/min and the column temperature was 35 °C.

Mass spectra were recorded in centroid mode in a mass range from m/z 60 to 1050 m/z.

The mass spectrometer operated using a capillary voltage of 3.7 kV. The source temperature was set to 285 °C, with 14 L/min drying gas and a nebulizer pressure of 45 psig. Fragmentor, skimmer, and octopole voltages were set to 175, 65, and 750 V, respectively.

Active reference mass correction was performed through a second nebulizer using the reference solution (m/z 112.9855 and 1033.9881) dissolved in mobile phase 2-propanol–acetonitrile–water (70:20:10 v/v). Data were acquired from m/z 60–1050. Data analysis and isotopic natural abundance correction were performed with MassHunter ProFinder (Agilent). Normalization was performed based on protein content.

2.6. NADP/NADPH Ratio and Malondialdehyde AssayNADP/NADPH ratio was tested using a dedicated Assay Kit (Abcam; Cat#ab65349), following the manufacturer’s instructions. Malondialdehyde (MDA) levels were evaluated by the thiobarbituric acid-reactive substance assay [18,19]. In all cases, enzymatic activity was normalized for total protein concentrations tested using Bradford analysis [20]. 2.7. Radioactive Glucose Analog Uptake and its Relationship with Glycolytic FluxIn vitro 18F-Fluoro-deoxyglucose (FDG) uptake was estimated using the LigandTracer White® instrument (Ridgeview, Uppsala, SE) according to our previously validated procedure [21,22,23]. Briefly, the device consists of a beta-emission detector and a rotating platform harboring a standard Petri dish. The rotation axis is inclined at 30° from the vertical so that the organ alternates its position from the nadir (for incubation) to the zenith (for counting) every minute.To calibrate the measured counting rate, a square of absorbent paper (5 mm × 5 mm) was stuck on the counted Petri point and soaked with FDG (200 to 600 KBq). Radioactivity decay was thus monitored for 12 h as shown in Supplementary Material Figure S1. The calibration factor was checked every week and was set by the slope of the regression line to 0.03 ± 0.0006. According to this process, the measured counting rate (in counts per second or cps) was thus divided by 0.03 to define the true radioactivity content in the tested sample.The same culture used for the Seahorse analysis was used to collect 10–15 × 103 cells of each type that were seeded and cultured for two days. Soon before the experiment, cells were immersed in 3 mL of the same culture medium containing 1.8 to 2.2 MBq/mL FDG. After 120 min monitoring, obtained counting rate was normalized according to the following equation:

FDG culture content=Culture counting rate (cps)calibration factor×administered radioactivity (Bq)×cell number

(1)

The conventional Sokoloff model of FDG uptake [24], was used to estimate the metabolic rate of glucose (MRGlu*) as the product of retained FDG fraction times the total available glucose (5.5 × 3 = 15 µMol) eventually divided for the observation period of 120 m.This estimation was thus divided by the direct measurement of glycolytic-related glucose disposal (MRGlu) provided by the Seahorse approach in the corresponding experimental condition and on the same day. The ratio MRGlu*/MRGlu also called lumped constant [24], was thus calculated. 2.8. Western Blot Analysis

Protein extraction was performed in RIPA buffer supplemented with a protease inhibitor cocktail. Western blot (WB) experiments were performed according to the standard procedure with the following antibodies: anti-hexose 6-phosphate dehydrogenase (H6PD, #ab170895, Abcam, Cambridge, UK), anti-glucose 6-phosphate dehydrogenase (G6PD, #ab124738, Abcam, Cambridge, UK), anti-glucose 6-phosphate transport (G6PT or SLC37A4, #ab80463, Abcam, Cambridge, UK), anti-glucose-6-phosphatase (G6Pase catalytic subunit, #ab83690, Abcam, Cambridge, UK), anti-mitofusin 2 (MFN2, #PA5-72811, ThermoFisher Scientific, Waltham, MA, USA), anti-dynamin-related protein 1 (#DRP1-101AP, ThermoFisher Scientific, Waltham, MA, USA), anti-translocase of the inner membrane (TIM) (TIM subunit 17A, #ab192246, Abcam, Cambridge, UK), anti-mitochondrial calcium uniporter (MCU, HPA016480 Sigma-Aldrich, and anti-β-actin (#15G5A11/E2, ThermoFisher Scientific, Waltham, MA, USA) or anti-HSP90 (BD Bioscience, BD610418 as the loading control. Specific secondary antibodies were employed (Sigma-Aldrich, St. Louis, MO, USA), all diluted 1:10,000 in PBS-Tween. Bands were detected and analyzed for optical density using an enhanced chemiluminescence substrate (ECL, Bio-Rad, Hercules, CA, USA), a chemiluminescence system (Alliance 6.7 WL 20M, UVITEC, Cambridge, UK), and UV1D software (UVITEC, Cambridge, UK). All the bands of interest were normalized with actin levels detected on the same membrane.

2.9. 2-NBDG/ER Colocalization2-[N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)amino]-2-deoxyglucose (2-NBDG) colocalization with the ER was performed by confocal microscopy as previously described [25]. Briefly, cells grown on 12 mm diameter glass coverslips were stained for 15 min with the fluorescent probes 2-NBDG (25 microM) and glibenclamide (0.5 mM) at 37 °C. After extensive washing, the cells were immediately imaged with an SP2-AOBS confocal microscope (Leica Microsystems, Mannheim, Germany). Four to eight randomly selected fields were analyzed in three independent samples for each treatment. Original unadjusted and uncorrected images were processed by Fiji software (Version 2.3, NIH). Colocalization was expressed as the percent of 2-NBDG-containing pixels that colocalize with glibenclamide-positive ones and glibenclamide-containing pixels that colocalize with 2-NBDG-positive ones. 2.10. Mitochondrial–ER Colocalization and Mitochondrial MorphologyCells were seeded onto glass coverslips (18 mm diameter for imaging experiments) and transfection was performed at 60% confluence using Lipofectamine 2000 (Invitrogen) with 1 µg of DNA. Split-GFP-based contact site sensors (SPLICSs) were used as previously described [26,27]. Images were collected on a Leica TCS SP5 II confocal system using a WLL white laser (Leica) equipped with a PlanApo 100× (numerical aperture 1.4 objective). The pinhole was set to 1 Airy unit and the imaging was performed at 1024 × 1024 pixels per image, with a 0.2 Hz acquisition rate. Photomultiplier gain was adjusted and maintained among different experiments to minimize background and avoid saturation. Images were elaborated with Fiji software (Version 2.3, NIH). Mitochondrial–ER colocalization: To count ER–mitochondria contacts, a complete z-stack of the cell was acquired every 0.42 µm. Z-stacks were processed using Fiji. Images were first convolved and filtered using the Gaussian blur filter. A 3D reconstruction of the resulting image was obtained using the Volume J plugin. Two selected faces of the 3D rendering were then thresholded and used to count ER–mitochondria contact sites. Mitochondrial morphology: cells expressing the 4mt-circular permuted Venus (cpV) were morphologically analyzed, as described previously [28], using the Fiji software. 2.11. Ca2+ Experiments

Aequorin experiments were carried out on a PerkinElmer EnVision plate reader equipped with a two-injector unit. Cells were seeded in 24-well plates and then re-plated into 96-well plates (1:5 dilution) the day before the experiment. After reconstitution with 5 μM coelenterazine, cells were placed in 70 μL of extracellular saline (in mM: 140 NaCl, 2.8 KCl, 2 MgCl2, 10 HEPES, 1 CaCl2, 5 glucose; pH 7.4 at 37 °C) and luminescence from each well was measured for 1 min. During the experiment, 100 µM ATP was first injected at the desired concentration to activate Ca2+ transients in 0.5 mM EGTA-containing extracellular saline in the absence of Ca2+, and then a hypotonic, Ca2+-rich, digitonin-containing solution (10 mM CaCl2 in H2O) was added to discharge the remaining aequorin pool. Output data were analyzed and calibrated with a custom-made macro-enabled Excel workbook.

2.12. Real-Time Polymerase Chain Reaction

Total RNA was purified using a Quick RNA Miniprep Plus kit (Zymo Research) according to the manufacturer’s instructions and reverse transcribed using the SensiFAST cDNA Synthesis KIT (Bioline, Aurogene). Real-time polymerase chain reaction (RT-PCR) was performed using SensiFAST SYBR (Bioline). Primers were designed using PRIMER 3 software and purchased from TIB MOLBIOL (Genoa, Italy). Primer sequences and RT-PCR conditions are available upon request.

The following gene levels were analyzed: isoforms 2 and 3 of sarco/endoplasmic reticulum Ca2+-ATPase (SERCA), ryanodine receptors (RYR), and isoforms 2 and 3 of inositol 1,4,5-trisphosphate receptors (IP3R). Target gene levels were normalized to that of glyceraldehyde-3-Phosphate dehydrogenase (GAPDH) and hypoxanthine–guanine phosophoribosyltransferase (HPRT) mRNA. The Q-Gene program was used for analyzing gene expression [29]. 2.13. Statistical Analysis

All experimental groups were studied in n ≥ 3. Data are presented as mean  ±  standard deviation (SD). Differences among the experimental conditions were tested using analysis of variance (ANOVA), as appropriate. Statistical significance was considered for p values  <  0.05. Statistical analyses were performed using SPSS software (Version 26, IBM).

4. DiscussionMost neoplastic lesions are known to facilitate the flux of G6P into the PPP, branching from glycolysis at the first committed step of glucose metabolism. This metabolic shift allows fueling the building blocks and the reductive power needed by high-rate anabolic processes and antioxidant responses. In addition to its role as the unique source of ribose-5P for ribonucleotides synthesis, the capability of PPP to preserve the intracellular NADPH pool [12,44,45,46,47] largely exceeds the contribution of glutamate dehydrogenase [48], malic enzyme 1 [49] and isocitrate dehydrogenases in the noncanonical TCA direction [50]. Classical assumptions consider PPP as a cytosolic glucose metabolism triggered by G6PD. Nevertheless, glutamine deprivation simultaneously decreased the proliferation rate of MDA cells, several PPP intermediates, and the NADPH/NADP ratio. Combined with the evident oxidative damage, these observations indicate a slowdown in PPP activity that mismatched the marked increase in G6PD expression, observed in our experiments in agreement with previous studies [13].This apparent paradox is explained by the opposite behavior of the PPP branch confined within the ER that, although scarcely considered, has been found to provide a high contribution to the ribose production and NADPH generation of several types of cancer cells [16,34]. Indeed, the expression of its triggering enzyme H6PD was decreased after glutamine shortage in both hormone-sensitive and triple-negative cultures. However, MDA cells coupled this response with an increased expression of G6Pase, able to prevent G6P access to H6PD due to the shared reticular confinement.The metabolic correlate of this enzymatic impairment under glutamine deprivation is corroborated by several observations. Indeed, the combination of an evident glutamine addiction with a relatively low G6PD abundance agrees with a high dependence of the rapidly proliferating MDA cells on the ER-PPP. Likewise, the consequence of H6PD downregulation on NADPH reductive power and the consequent redox damage was particularly pronounced in these triple cells. Finally, the same MDA cells showed higher rates of both FDG uptake and reticular accumulation of its fluorescent counterpart 2-NBDG with respect to their hormone-sensitive counterparts with both indexes particularly impaired by glutamine shortage. A large body of evidence previously documented that, once phosphorylated by hexokinases, these two glucose analogs cannot be channeled to glycolysis and cytosolic PPP, being false substrates for phosphoglucose isomerase and G6PD, respectively [51]. By contrast, they can be—and indeed are—processed by the “omnivore” H6PD enzyme, whose catalytic function prevents their hydrolysis by G6Pase and the subsequent glucose transporter (GLUT)-mediated release to the cytosol. Accordingly, the agreement between tracer kinetics, enzymatic asset, and intermediate levels configures the deceleration of G6P flux through the ER-PPP as a primary determinant of the NADPH reductive power in cancer.According to our protocol, implying partial amino acid deprivation, glutamine starvation was particularly severe in the aggressive model of MDA cultures that are relatively depleted of glutamine synthase as opposed to the preserved enzyme expression of MCF7 ones [52]. In agreement with this experimental design, glutamine shortage did not affect the culture growth of these hormone-sensitive cells, suggesting a direct dependence of MDA anabolic pathways on the glutamine excess typical of the in vitro cultures. A similar consideration also applies to the behavior of MRGlu and OCR, whose slowdown in the triple-negative breast cancer cells was associated with an increased ATP/AMP ratio. This response suggests an adaptation of energy-producing pathways to a decreased metabolic demand. More importantly, it also confirms that the high glycolytic flux typical of the Warburg effect is largely independent of energy needs.On the other hand, the impairment in maximal respiratory capacity induced by glutamine deprivation confirms that this amino acid contributes to the regulation of mitochondrial activity, at least in triple-negative cells. This role has most often been attributed to the capability of glutamine-derived glutamate to supply the Krebs cycle with the nitrogen and carbon equivalents needed for its anaplerotic function, threatened by the high citrate efflux to the cytosol for fatty acid synthesis. However, these same data document a relevant role for ER–mitochondria networking in this setting. Indeed, the decreased number of ER–mitochondria juxtapositions induced by glutamine shortage were paralleled by MCU upregulation, increased mitochondrial Ca2+ uptake, and oxidative stress as a preliminary step for the downregulation of the MDA proliferation rate [53]. This finding reproduces previous observations about the mandatory role of H6PD function in preserving the ER–mitochondria networking, the cellular redox state, and thus the mitochondrial Ca2+ content. Similarly, it fits with the high levels of glutamine transporters that have been found to characterize the MAMs [54]. Accordingly, these data seem to suggest that glutamine addiction of rapidly proliferating cancer cells might involve the ER-PPP as a powerful, so far unrecognized, regulator of cancer metabolic reprogramming through its capability to modulate the mitochondrial function. 5. Conclusions

In conclusion, the present study indicates that ER-PPP contributes to the glutamine dependence of anabolic processes at least in the selected model of aggressive triple-negative breast cancer. Its reticular localization configures this pathway as a fundamental regulator of signals triggering cell proliferation through the ER–mitochondria connection.

Although the underlying mechanisms cannot be defined, the opposite response of H6PD and G6PD expression to glutamine shortage indicates profound independence of these two geographically unconnected pathways. Likewise, the agreement of the ER-PPP rate with the proliferating activity configures this reticular metabolism as a primary determinant of cancer aggressiveness.

This observation might open future windows for the therapeutic targeting of cancer metabolism. So far, they provide an alternative and robust explanation of the high diagnostic and prognostic power of 18F-FDG uptake. This clinical description of cancer aggressiveness would be independent of glucose consumption, the most elementary feature of life, shared by all living cells. Rather, it would derive from the capability of this tracer to describe the activity of a specific ER metabolism strictly connected with cancer growth.