Glucose withdrawal induces supra-physiological levels of tyrosine phosphorylation in cells sensitive to glucose withdrawal. (A) LN18, LN229, T98, and U87 GBM cell lines were starved of glucose and pyruvate for 24 h. Trypan blue exclusion measurements demonstrated that LN18, T98, and U87, but not LN229, show a rapid and complete loss of viability following glucose withdrawal. (B) Western blotting with a phospho-tyrosine antibody revealed that the glucose withdrawal-sensitive cell lines (LN18, T98, and U87) exhibit a dramatic induction of phospho-tyrosine signaling at the indicated times following glucose withdrawal. Conversely, the glucose withdrawal-insensitive cell line LN229 showed no increase in phospho-tyrosine levels following glucose withdrawal. Actin served as an equal loading control. (C–E) Glucose withdrawal induces supra-physiological levels of phospho-tyrosine signaling even in cells expressing a constitutively active tyrosine kinase. (C) Expression of the constitutively active EGFR mutant EGFRvIII in U87 cells induces high levels of phospho-tyrosine signaling as demonstrated by western blotting with an anti-phospho-tyrosine antibody. (D) Expression of the constitutively active EGFRvIII mutant in U87 cells (U87-EGFRvIII) does not alter the sensitivity to glucose withdrawal at 24 h. Error bars are standard deviation of the mean. (E) U87-EGFRvIII cells demonstrate a rapid induction of phospho-tyrosine signaling at the times indicated following glucose withdrawal. For phospho-tyrosine, both dark and light film exposures are shown. (F–I) Human sarcoma- and melanoma-derived cell lines also demonstrate a correlation between sensitivity to glucose withdrawal and rapid induction of supra-physiological phospho-tyrosine levels. (F) The sarcoma cell lines HT161 and TC32 were starved of glucose and pyruvate and their viability was measured by Trypan blue exclusion at the indicated times. TC32 cells show a rapid and complete loss of viability, whereas HT161 were relatively insensitive to glucose withdrawal. Error bars are standard deviation of the mean. (G) Western blotting with an anti-phospho-tyrosine antibody demonstrated that TC32, but not HT161, show induction of hyper-phosphorylation after 2 h of glucose starvation. (H) Melanoma cell lines M202, M207, M229, and M249 were starved of glucose and pyruvate for 24 h. Trypan blue exclusion measurements revealed that M202 and M207, but not M229 or M249 cells, showed significant loss of cell viability after 24 h of glucose withdrawal. Error bars are standard deviation of the mean. (I) Western blotting demonstrated that 7 h of glucose withdrawal induced supra-physiological tyrosine phosphorylation in M202 and M207, but not M229 or M249.

PTEN status regulates glucose withdrawal-induced phospho-tyrosine induction and cell death. (A, B) Murine wild-type Pten or the lipid phosphatase inactive mutants G129E and C142S were expressed in the PTEN-null GBM cell line U87. (A) U87 cells infected with an empty vector control, wild-type Pten or the lipid phosphatase inactive mutants were starved of glucose and pyruvate for 16 and 24 h, and viability was measured by Trypan blue exclusion. Expression of wild-type Pten increased survival following glucose withdrawal by roughly two-fold (P=0.04 (*) and 0.001 (**) by Student's t-test). Error bars are standard error of the mean (n=3). (B) Western blotting with an anti-phospho-tyrosine antibody demonstrated that U87-Pten cells exhibit reduced induction of phospho-tyrosine signaling after 3 h of glucose withdrawal. Pten and GRB2 served as confirmation of Pten overexpression and equal loading, respectively. (C, D) HCT116 cells with genetic knockout of PTEN show greater cell death and phospho-tyrosine induction in response to glucose withdrawal. (C) HCT116 with wild-type PTEN (WT) or genetic knockout by homologous recombination (HCT116-22, KO) were serum starved for 16 h and then starved of glucose and pyruvate for 24 h before measurement of viability by Trypan blue exclusion. PTEN knockout cells showed reduced survival at 24 h compared with wild-type cells (P-value=0.03 (*) by Student's t-test, n=4). (D) Western blotting with an antibody against phospho-tyrosine demonstrated increased phospho-tyrosine signaling in HCT116 PTEN KO cells following 3 h of glucose withdrawal. Confirmation of loss of PTEN expression in HCT116 KO cells is shown in Supplementary Figure S4. (E, F) RWPE cells with reduced PTEN expression show greater cell death in response to glucose withdrawal (P-value=0.002 (*) by Student's t-test, n=3). (E) RWPE cells with wild-type PTEN or cells infected with an shRNA targeting PTEN were starved of glucose and pyruvate for 24 h before viability measurements by Trypan blue exclusion. Confirmation of reduced PTEN expression is shown in Supplementary Figure S4. (F) Western blotting demonstrated that RWPE cells with reduced PTEN expression demonstrate greater induction of phospho-tyrosine signaling following glucose withdrawal.

Phospho-proteomics reveals that glucose withdrawal induces a distinct signature of phospho-tyrosine signaling that is associated with focal adhesions. (A, B) Following glucose and pyruvate starvation of U87 and U87-EGFRvIII for the indicated times, western blotting revealed activation of some but not all signaling pathways following glucose withdrawal. (A) Phospho-specific antibodies against tyrosine residues demonstrated significantly increased phosphorylation of RTKs, including EGFR, Met, and PDFGRβ. The non-RTK Src also showed increased active site phosphorylation. Total EGFR and actin served as equal loading controls. (B) Phospho-specific antibodies revealed increased glucose withdrawal-induced activity of all MAPK pathways tested (ppERK 1/2, pJNK and p-p38α) but not mTOR signaling (pS235/S236-S6) or Akt signaling (pS473-Akt). (C–E) Glucose withdrawal induces a signature of hyper-phosphorylation in U87 that is associated with focal adhesions. (C) Hierarchical clustering of tyrosine phosphorylation in U87 cells reveals that glucose withdrawal induces a distinct set of phospho-events. U87 cells were treated with four stimuli known to induce tyrosine phosphorylation, including (a) EGF stimulation (10 ng/ml, 5 min), (b) vanadate treatment (1 mM, 60 min), (c) H₂O₂ (5 mM, 30 min) and (d) glucose and pyruvate withdrawal (3 h). Changes in phospho-tyrosine signaling were measured by quantitative, label-free mass spectrometry (Rubbi et al, 2011) and data were hierarchically clustered. Each row of the heatmap depicts an individual phosphorylation event, and each column represents a sample as labeled. In the heatmap, red and green represent normalized levels of high and low phosphorylation, respectively. Samples were measured in technical duplicate (r1 and r2). Branches of the dendrogram associated with upregulation by glucose withdrawal and vanadate treatment are colored orange and blue, respectively. See Supplementary Table 1 for quantitative phospho-peptide data. (D) Glucose withdrawal induces increased phosphorylation of proteins known to localize to focal adhesions. Tyrosine residues on integrin β 1 (ITGB1 pY783), caveolin 1 (CAV1 pY14), and ephrin 2A (EPH2A pY588 and pY594) show dramatically increased phosphorylation in response to 3 h of glucose withdrawal. (E) Phospho-events associated with focal adhesions are enriched following glucose withdrawal in U87 cells. Phospho-peptides were ranked according to the measured log₂ fold change in phospho-tyrosine levels following glucose withdrawal and plotted on a waterfall plot, where red and green represent increased or decreased phosphorylation, respectively. Analysis of the phospho-peptides demonstrated an enrichment for proteins annotated with the GO term Focal Adhesion (GO:0005925) at the top of the ranked list (i.e., increased phosphorylation following glucose withdrawal) (permutation-based P-value=0.02). (F) Western blotting revealed increased FAK Y397 phosphorylation in response to glucose withdrawal. Source data is available for this figure in the Supplementary Information.

Glucose withdrawal induces rapid amplification of ROS in glucose withdrawal-sensitive cells. Cells were starved of glucose and pyruvate for 0 or 3 h, stained with either the oxidation-sensitive fluorogen DCF-DA or mitoSOX, and analyzed by flow cytometry. (A, B) LN18 but not LN229 cells demonstrated increased DCF-DA fluorescence 3 h following glucose and pyruvate starvation. (C) Quantification of the fold change in DCF-DA signal (3 h:0 h) revealed that the glucose withdrawal-sensitive cell lines (LN18, T98, and U87) but not glucose withdrawal-insensitive cell lines (LN229 and M229) demonstrated increased ROS levels following glucose and pyruvate withdrawal. Histograms for all cell lines are shown in Supplementary Figure S7. (D, E) LN18 but not LN229 cells demonstrated increased mitochondrial superoxide production 3 h after glucose and pyruvate starvation. (F) Quantification of the fold change in mean mitoSOX signal (3 h:0 h) demonstrated that glucose withdrawal-sensitive cell lines (LN18, T98, and U87) but not glucose withdrawal-insensitive cell lines (LN229 and M229) demonstrated increased levels of mitochondrial ROS following glucose and pyruvate withdrawal. Histograms for all cell lines are shown in Supplementary Figure S7.

Glucose withdrawal-induced ROS mediate tyrosine kinase induction and cell death. (A) U87, LN18, and LN229 cells were starved of glucose and pyruvate for the indicated times with or without the H₂O₂ scavenger catalase (250 U/ml for U87, 1 kU/ml for LN18 and LN229). Western blotting demonstrated that glucose withdrawal-induced phospho-tyrosine signaling was ablated by catalase treatment in U87 and LN18. In addition, glucose withdrawal-induced activation of the Src active site (Y416) in LN18 cells required catalase-sensitive ROS. Actin served as an equal loading control. (B) U87-EGFRvIII cells were starved of glucose and pyruvate for 3 h in the presence of either DMSO or the antioxidant MnTMPyP (25 μM). Western blotting revealed that MnTMPyP treatment reduced tyrosine phosphorylation of EGFRvIII (∼155 kDa, the most prominent band in the phospho-tyrosine western blot) following glucose withdrawal. (C, D) ROS are required for cell death following glucose withdrawal. (C) The glucose withdrawal-sensitive cell lines LN18, U87, and U87-EGFRvIII were starved of glucose and pyruvate with or without catalase, and viability was measured by Trypan blue exclusion 24 h later. Catalase treatment rescued cells from glucose withdrawal-induced cell death (P-value <1 × 10⁻³ (*) by Student's t-test). Error bars are standard deviation of the mean (n=3–13). (D) U87-EGFRvIII cells were starved of glucose and pyruvate for 24 h in the absence or presence of the antioxidant MnTMPyP. Treatment with MnTMPyP rescued U87-EGFRvIII from glucose withdrawal-induced cell death (P-value <1 × 10⁻³ (*) by Student's t-test, n=13 and 3 for No treatment and MnTMPyP treatment, respectively).

NADPH oxidase- and mitochondria-derived ROS contribute to glucose withdrawal-induced phospho-tyrosine signaling. (A) Inhibition of NOX inhibits glucose withdrawal-induced signaling. LN18, T98, and U87 cells were starved of glucose and pyruvate in the presence of DMSO or DPI (1 μM). Western blotting demonstrated that NOX activity is required for the induction of phospho-tyrosine signaling. (B) Knockdown of the NOX subunit p22^phox attenuates phospho-tyrosine signaling following glucose withdrawal. U87 cells were reverse transfected with control, non-targeting siRNA or siRNA against p22^phox, DUOX1 or DUOX2. Forty-eight hours later, cells were starved of glucose and pyruvate for 5 h. Western blotting demonstrated that knockdown of p22^phox but not DUOX1/2 attenuated glucose withdrawal-induced phospho-tyrosine signaling. p22^phox knockdown efficiency was >90% (Supplementary Figure S9). (C) LN18, T98, and U87 cells were starved of glucose and pyruvate in the presence of either DMSO or BAPTA-AM (25 μM). Western blotting with an anti-phospho-tyrosine antibody demonstrated that chelation of intracellular Ca²⁺ by BAPTA-AM completely abrogated glucose withdrawal-induced phospho-tyrosine signaling. Treatment with extracellular EDTA (25 μM) had no effect. Actin and GRB2 served as equal loading controls. (D) ρ₀ derivatives of T98 and U87 cells do not exhibit upregulation of phospho-tyrosine signaling or activation of Src in response to glucose withdrawal. (E) Catalase rescues parental but not ρ₀ cells from glucose withdrawal-induced cell death. Cells were starved of glucose and pyruvate with or without catalase (1 kU/ml), and viability was measured by Trypan blue exclusion 24 h later.

Glucose withdrawal-induced ROS mediate oxidative inhibition of protein tyrosine phosphatases. (A) Glucose withdrawal causes oxidative inhibition of PTP activity in glucose withdrawal-sensitive cells. U87 and LN229 cells were starved of glucose for 0 or 3 h, and the ability of cell lysates to dephosphorylate a phospho-substrate was measured by quantitative western blotting. Addition of DTT to the dephosphorylation reaction reduced oxidized PTPs. Data were normalized to a dephosphorylation reaction with the PTP inhibitor vanadate added (P=0.04 (*) and 0.01 (**) by Student's t-test (n=3 for U87). A representative blot is shown in Supplementary Figure S12A. (B) Glucose withdrawal inhibits PTP-1B activity by oxidation. U87 and U87-EGFRvIII cells were starved of glucose and pyruvate for 1.5 h with or without catalase. PTP-1B was then immunoprecipitated from cell lysates under anaerobic conditions and incubated with the colorimetric phosphatase substrate pNPP. PTP-1B activity was normalized to a no-antibody control and expressed in arbitrary units (AU). For both cell lines, glucose withdrawal reduced PTP-1B activity by two-fold and catalase treatment rescued PTP-1B activity from the effects of glucose withdrawal (P<0.01 (*) and 0.02 (**) by combined Fisher's method of paired Student's t-test for U87 and U87-EGFRvIII, n=4 for U87, n=8 for U87-EGFRvIII). (C) Glucose and pyruvate starvation and vanadate treatment synergistically kill U87 cells. U87 cells were exposed to increasing doses of vanadate for 1.5 h and then starved of glucose and pyruvate for the indicated times. The number of viable cells was measured by Trypan blue exclusion 24 h later. To assess the degree of synergy, the combination index (Comb. Index) was calculated using the method of Chou and Talalay (1984). Values less than one indicate positive synergy. Error bars are the standard deviation of the mean (n=2). (D) Isobologram plot of the effect of glucose and pyruvate starvation combined with vanadate treatment demonstrates positive synergy. The effective doses (ED) of glucose and pyruvate starvation (min) and vanadate treatment (mM) are plotted on the x and y axis, respectively. Lines of linear additivity connect the ED for ED₅₀, ED₇₅, and ED₁₀₀ for individual treatments. Because the experimentally measured responses to combinations of glucose starvation and vanadate treatment (combination treatments) lie to the left of the linear lines of additivity, glucose starvation and vanadate treatment interact with positive synergy. Source data is available for this figure in the Supplementary Information.

Glucose withdrawal activates a positive feedback loop resulting in supra-physiological phospho-tyrosine signaling and ROS-mediated cell death. In cells dependent on glucose for survival, glucose and pyruvate deprivation induces oxidative stress driven by NOX and mitochondria. This oxidative stress provokes a positive feedback loop in which NOX and mitochondria generate superoxide anion (O₂⁻), which dismutes to hydrogen peroxide (H₂O₂) and inhibits PTPs by oxidation (e.g., PTP-1B and PTEN). Without the negative regulation of PTPs, TKs including EGFR and Src activate NOX at focal adhesions, further amplifying ROS generation. This glucose withdrawal-induced positive feedback loop results in supra-physiological levels of tyrosine phosphorylation and ROS-mediated cell death.