MCF-7 (human hormone sensitive breast cancer cell line) was purchased from ATCC (American Type Culture Collection). The cell line was maintained at 37 °C with 5% CO${}_2$, validated by short tandem repeat (STR) profiling and tested negative for mycoplasma contamination using Myco-Alert Mycoplasma Detection Kit (#21793-0127, Lonza, Basel, Swizzerland).

MCF-7 cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM) (#01-057-1A, Sartorius, AG, Göttingen, Germany) supplemented with 10% fetal bovine serum (#10270-106, Thermo Fisher Scientific , Waltham, MA, USA), 25 mM glucose, 2 mM glutamine, 1 mM pyruvate, 1% penicillin-streptomycin (control medium) (Thermo Fisher Scientific, Waltham, MA, USA).

Cells were seeded in DMEM medium and after five hours the cell cultures were exposed to either fresh control medium (CTR) or DMEM containing only 2.5 mM glucose and 0.5 mM glutamine (LB). After 48 h the Cells were harvested for measurements.

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

Cell proliferation rate was evaluated using carboxyfluorescein-diacetate-succinimidyl ester (CFSE) (Merck KGaA, Darmstadt, Germany). Briefly, cells were plated in 6-well plates and after 24 h later the cells were labeled with CFSE (0.3 µM in PBS). After 20 min at 37 °C the labeling solution was replaced with a fresh culture medium, cells were incubated for another 20 min at 37 °C and then cultured either with CTR or LB medium for 48 h. CFSE fluorescence was acquired on a FACScan flow cytometer and the data were analyzed by FlowJo software (Version 10.8.1, Becton–Dickinson, Eysin, Swizzerland).

OCR and ECAR were measured using Seahorse XFp Extracellular Flux Analyzer (Agilent Technologies, Santa Clara, CA, USA) following manufacturer’s istructions. 7000 cells per well were plated in XFp plates and incubated in control medium. The medium was replaced 5 h later and the cells were incubated with either CTR or LB medium for 48 h. Briefly, cells were incubated in Agilent Seahorse DMEM, pH 7.4 (#103575, Agilent Technologies, Santa Clara, CA, USA), enriched with nutrients mimicking the respective incubation medium conditions. OCR and ECAR were measured at 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) and the ATP synthesis uncoupler carbonyl cyanide-4-trifluoromethoxyphenylhydrazone (FCCP, 1.25 µM).

OCR was expressed as picomol O${}_2$ $\times$ min${}^{-1}$/million cells while ECAR was turned into proton efflux rate (PER) using WAVE software (Version 2.4, Agilent Technologies, Santa Clara, CA, USA) after mesuring the buffer factor of the control and LB medium. Glucose consumpions (nanomoles) through glycolysis was calculated converting PER value expressed in picomol H${}^{+}$$\times$ min${}^{-1}$/million cells to glucose nanomoles $\times$ min${}^{-1}$/million cells, applying the equation Glucose + 2 ADP + 2 Pi –$>$ 2 Lactate + 2 ATP + 2 H${}_2$O + 2 H${}^{+}$.

Lactate and pyruvate in cell lysates were measured in a spectrophotometric assay following the reduction of NAD${}^{+}$ at 340 nm. Data were normalized for total protein concentrations [10].

Cells were lysed in RIPA buffer (#9806, Cell Signaling Technologies, Danvers, MA, USA) plus protease inhibitor cocktail (G135, Applide Biological Materials, Richmond, Canada). Western blot experiments were performed according to the standard procedure. Primary antibodies employed are the following: anti-hexose 6-phosphate dehydrogenase (H6PD, #ab170895, Abcam, Cambridge, UK), anti-glucose 6-phosphate dehydrogenase (G6PD, #ab124738, Abcam, Cambridge, UK) anti- glucose regulated protein 78 (GRP78, #ab21685, Abcam, Cambridge, UK) and anti-Actin (#MA1-140, ThermoFisher Scientific, Waltham, MA, USA). Relevant secondary antibodies (Sigma-Aldrich, St. Louis, MO, USA) were, diluted 1:10,000 in PBS-Tween, developed with chemiluminescence substrate (#170-5060, Bio-Rad, Hercules, CA, USA). Detection and densitometric analyses were performed using Alliance 6.7 WL 20M and UVITEC software (Alliance-1D software, UVITEC, Cambridge, UK), respectively. All the bands of interest were normalized with actin levels detected on the same membrane.

Cells were incubated for 20 min at 37 °C with 1 µM ER-Tracker Red, 5 µM H${}_2$DCFDA or 5 µM MitoSOX (Invitrogen by Thermo Fisher Scientific, Waltham, MA, USA). After two washes with PBS the cells were resuspended in PBS + 1% bovine serum albumin (BSA, Sigma-Aldrich, St. Louis, MO, USA) for flow cytometric analysis (FACSCan, Becton Dickinson, Eysin, Swizzerland). Data analysis was performed with FlowJo software (Version 10.8.1, Becton–Dickinson, Eysin, Swizzerland).

NADPH/NADP and NADH/NAD${}^{+}$ ratio was valuated using a dedicated Assay Kit (ab65349, Abcam, Cambridge, UK), following the manufacturer’s instructions. Data were normalized for total protein concentrations.

In vitro ${}^{18}$F-Fluoro-deoxyglucose (FDG) uptake was estimated using the LigandTracer White® instrument (Ridgeview, Uppsala, Sweden) according to our previously validated procedure [10, 11, 12]. The same culture used for the Seahorse analysis was used to collect 10–15 $\times$ 10${}^4$ MCF-7 cells that were seeded and cultured for two days. Right before Ligand Tracer measurements, cells were incubated with 3 mL FDG conditioned medium (concentration range 1.8 to 2.2 MBq/mL). Obtained counting rate (after 120 minutes) was normalized using the following equation:

$$\begin{array}{c}\hfill \mathit{\text{FDG culture content}}=\frac{\mathit{\text{Culture counting rate (cps)}}}{\mathit{\text{calibration factor}}\times \mathit{\text{administered radioactivity (Bq)}}\times \mathit{\text{cell number}}}\hfill \end{array}$$

Lumped constant (LC) was calculated as described before [13]. Briefly, LC is obtained as the ratio between the metabolic rate of glucose (MRGlu*) and glycolytic-related glucose disposal (MRGlu). Were MRGlu* is defined as the product of retained FDG fraction and the total available glucose (5 $\times$ 3 = 15 µMol) divided by 120 minutes. While MRGlu was directly measured by Seahorse Analyzer in a twin culture following the same experimental conditions.

Data are presented as mean $\pm$ standard deviation (SD), n $\ge$ 3. Unpaired t test or analysis of variance (ANOVA) were used to test the differences among the experimental conditions. Statistical significance was considered for p values 0.05. GraphPad Prism 9.3.1 (GraphPad, San Diego, CA, USA) were used for statistical analyses.

Prolonged (48 hours) glucose-glutamine shortage slightly, yet significantly, decreased the growth curve of MCF-7 cells (Fig. 1A). This effect was not related to any increase in mortality rate, since propidium iodide negative cells remained invariant during nutrient stress (Fig. 1B, Supplementary Fig. 1A). By contrast, it reflected a reduced MCF-7 cells proliferation rate, as documented by the CSFE dilution (Fig. 1C, Supplementary Fig. 1B).

Fig. 1.

Cell proliferation rate and metabolic pattern of MCF-7 under prolonged starvation. (A–C) Cell numbers (A), percent of viable cells (B) and CFSE mean fluorescence intensity (MFI, C) of MCF-7 cultures under CTR (dark green) or LB (light green) medium. (D–G) Glycolytic flux (D), baseline OCR (Non-ATP-linked (solid columns) and ATP-linked (white columns), E), OCR in response to FCCP (F) and NADH/NAD${}^{+}$ ratio (G) of CTR (dark green) or LB (light green) cultures. (H) OCR under CTR (dark green) or LB (light green) condition in the absence or presence of BPTES expressed as a percent of untreated cells: Non-ATP-linked (solid columns) and ATP-dependent (dashed columns). (I,J) BPTES-untreated ratio of intracellular pyruvate (I) and lactate content (J) in MCF-7 cultured in CTR (dark green) or LB (light green) medium. Data are represented as mean $\pm$ SD. n $\ge$ 3, experiments per group. ns = not significant; * = p 0.05; ** = p 0.01. CFSE, carboxyfluorescein-diacetate-succinimidyl ester; CTR, control medium; LB, glucose-glutamine shortage; OCR, oxygen consumption rate; ATP, adenosine triphosphate; FCCP, carbonyl cyanide-4-trifluoromethoxyphenylhydrazone; BPTES, Bis-2-(5-phenylacetamido-1,3,4-thiadiazol-2-yl)-ethyl sulfide.

The metabolic reaction to energy substrate depletion implied a different response by glycolysis and cell respiration. As expected, the ten-fold shortening of glucose availability decreased lactate release. Nevertheless, PER slowed down by only 30% (Fig. 1D), apparently suggesting a metabolic shift increasing the glucose fraction channeled to the glycolysis. Baseline OCR was apparently not affected by the prolonged nutrient deprivation, although its ATP-synthase dependent fraction was higher in starved than in control cultures (Fig. 1E). This respiratory reaction agreed with the relatively preserved response to FCCP-induced uncoupling (Fig. 1F) and matched a slight and not significant increase in NADH/NAD${}^{+}$ ratio (Fig. 1G), suggesting a preserved capability of mitochondria to increase their OXPHOS contribution to the cell energy asset. Yet, it also mismatched the preferential channeling of glucose toward its incomplete conversion to lactate.

We thus hypothesized a possible enhancement of glutamine channeling to the Krebs cycle. To confirm this hypothesis, we evaluated the response of OCR and PER to the inhibition of glutamine conversion into glutamate. BPTES did not influence lactate release (data not shown) regardless of the nutrient availability. By contrast, the decrease in overall OCR induced by the inhibition of glutamine conversion to glutamate reflected two different mechanisms in the two experimental models: a selective lessening of oligomycin-insensitive OCR in control cultures as opposed to a selective slowing down of oxygen usage dedicated to ATP regeneration in starved ones (Fig. 1H).

This finding thus suggested that glutamine-derived carbons contribute to energy assets during prolonged nutrient deprivation, while their contribution to ATP synthesis is less represented under control conditions. This concept was at least partially confirmed by the estimation of intracellular levels of pyruvate and lactate. Indeed, cell content of both metabolites comparably decreased under starvation (Fig. 1I,J), suggesting that the incomplete glucose degradation was not caused by any increase in the catalytic function of LDH. On the other hand, BPTES administration decreased pyruvate levels to a greater extent with respect to lactate, suggesting a possible differential fate of glutamine-derived pyruvate equivalents compared to glycolysis-generated ones.

Altogether, these data suggested that the combined shortage of glucose and glutamine enhanced the incomplete degradation of glucose to lactate, apparently indicating a profound decrease in glucose utilization through the PPP. We thus aimed to verify whether this response was caused by an impairment of the catalytic function of either G6PD or H6PD. The abundance of the house-keeping protein ß-Actin was not affected by starvation (Fig. 2A,B,D and Supplementary Fig. 2). G6PD expression was slightly—though not significantly—increased after starvation (Fig. 2A). By contrast, its reticular counterpart H6PD was markedly decreased (Fig. 2B). This response was not the consequence of an ER shrinkage since fluorescence intensity after incubation with ER-Tracker was superimposable in both experimental models (Fig. 2C, Supplementary Fig. 3A). Similarly, it was not associated with a significant stress, since GRP78 abundance was similar in both control and starved cultures (Fig. 2D).

Fig. 2.

Effect of glucose-glutamine depletion on redox homeostasis. (A,B) Relative optical density (ROD) (left) and a representative Western blot (right) of G6PD (A) and H6PD (B) normalized versus $\beta$-Actin, in cells grown under CTR (dark green) or LB (light green) condition. (C) MFI of ER-Tracker in MCF-7 cultured in CTR (dark green) or LB (light green) medium. (D) Densitometric analyses (left) and a representative Western blot (right) of GRP78 normalized versus $\beta$-Actin in cells grown under CTR or LB condition. (E–H) H${}_2$DCFDA (E), MitoSOX (F), MitoTracker (G) and NADPH/NADP ratio (H) in MCF-7 cultured in CTR (dark green) or LB (light green) medium. Data are represented as mean $\pm$ SD. n $\ge$ 3, experiments per group. ns = not significant; * = p 0.05; **** = p 0.0001.

We thus hypothesized that the decreased fuel to PPP, combined with the decreased expression of H6PD might have enhanced the oxidative stress of starved cells. Nevertheless, cellular and mitochondrial oxidative status remained unchanged or even decreased, as described by the DCF (Fig. 2E, Supplementary Fig. 3B) and MitoSOX staining (Fig. 2F, Supplementary Fig. 3C). On the other hand, MitoTracker staining at least partially reproduced the analysis of OCR, and documented a relative integrity of mitochondrial membrane polarization (Fig. 2G, Supplementary Fig. 3D). This behavior was coherent with the response of the NADPH/NADP ratio that even slightly increased after prolonged nutrient depletion (Fig. 2H). Accordingly, these data suggested that the combined shortage of glucose and glutamine eventually resulted in a reduced ROS generation.

According to the previous analyses, the combined glucose and glutamine shortage also decreased the ROS generation in the whole cell and in the mitochondria, respectively. We thus aimed to verify whether this response also characterized the reticular lumen, modifying the activity of the local PPP. To this purpose, we exploited the strict link of FDG uptake with the expression and the catalytic function of H6PD. As expected, FDG uptake showed a linear increase under both experimental conditions (Fig. 3A). Nevertheless, despite the marked decrease in the concentration of the competing unlabeled glucose, the rate of tracer retention was significantly lower in starved cells than in control ones (Fig. 3A). Tracer-based assessment of glucose consumption thus provided a divergent picture with respect to the direct measurement of hexose disappearance from the incubation medium. Indeed, the estimated MRGlu* was superimposable to the directly estimated MRGlu under control condition (Fig. 3B). By contrast, this proportionality was virtually lost in starved cultures (Fig. 3B). As a consequence, the selective impairment of FDG uptake induced by nutrient deprivation markedly decreased the lumped constant (the ratio MRGlu*/MRGlu) (Fig. 3C).

Fig. 3.

Endoplasmic reticulum (ER) metabolism under a nutrient shortage. (A) ${}^{18}$F-Fluoro-deoxyglucose (FDG) kinetic uptake (fraction of administered dose) in cells grown under CTR (dark green) or LB (light green) condition. (B) Direct estimation (solid columns) or FDG-based estimation (dashed columns) of glucose consumption in CTR (dark green) or LB (light green) culture. (C) The experimental lumped constant value of MCF-7 cultured in CTR (dark green) or LB (light green) medium. Data are represented as mean $\pm$ SD. n $\ge$ 3, experiments per group. * = p 0.05; ** = p 0.01, *** = p 0.001 versus CTR. ns = not significant; # = p 0.05, FDG-based estimation versus direct estimation of glucose consumption (B).