Fig 1.
Pathways and reactions included in ENGRO1. A crossed circle symbol indicates exchange reactions.
Fig 2.
Ensembles of different metabolic responses to altered boundary conditions.
Different uptake fluxes are imposed for glucose and glutamine within the range [0, 20]. While varying one parameter, the availability of the other is left at its baseline value (G:20 mM h-1; Q:20 mM h-1; O2:20 mM h-1). For each parameterization, the set of 50,000 multi-weighted objective functions is optimized. For the total sample (panel A) and for the different identified subsets (panels B-G), the average flux value of aconitase (green), biomass synthesis (blue), lactate secretion (magenta) and oxygen uptake (cyan) is reported as a function of glutamine (left or top insets) or glucose (right or bottom insets).
Fig 3.
ENGRO1 and NIH-Ras mouse fibroblast sensitivity to glutamine limitation.
A) Sensitivity of Ensemble E models to a reduction in glutamine availability from 20 to 2.5 mM h-1. Red arrows/boxes correspond to statistically significant increased fluxes/metabolites (p-value<0.05), blue arrows to decreased fluxes/metabolites; dashed-grey arrows to non-significant variations. Variations in fluxes always refer to the forward net flux. The arrow thickness is proportional to the Z-score. For the ratio LACT/Glc (production/consumption) average and standard deviations are reported. (B) Proliferation curve of NIH-Ras mouse fibroblasts grown in 4 mM Gln (HQ) and 0.5 mM Gln (LQ). The cells were collected and counted at the indicated time points. The glucose and lactate absolute quantifications in spent media, and the ratio between the two, were performed by GC-MS. Lactate dehydrogenase (LDH) enzyme activity was measured by enzymatic assay. (C) Intracellular relative metabolite abundances of AcCoA, LACT, Glc, Pyr, Mal, Succ, Fum measured by GC/MS. Pyruvate dehydrogenase (PDH), succinate dehydrogenase (SDH) and isocitrate dehydrogenase (IDH2) enzyme activity. To compare more easily the magnitude of change induced by the low glutamine condition, the concentration of each metabolite in the high glutamine condition is always taken as 1.0, regardless of it absolute value. (D) Relative metabolite abundance of non-essential amino acids as analyzed by GC/MS. Intracellular total ROS levels measured by using 5 mM DCFH2-DA staining. Mitochondrial ROS levels measured by MitoSOX Red mitochondrial superoxide indicator. To compare more easily the magnitude of change induced by the low glutamine condition, the concentration of each metabolite in the high glutamine condition is always taken as 1.0, regardless of it absolute value. All the experiments were performed on NIH-Ras grown in 4 mM Gln and 0.5 mM for 144 h. Error bars indicate the standard deviations (n = 5).
Fig 4.
ENGRO1 biomass optimization for different boundary conditions.
A) ENGRO1 maximal growth rate (z-axis and coloring) as a function of glucose (G) and oxygen availability when glutamine availability is constant (Q:1). The arrows represent example mutational paths in the fitness landscape. The insets represent 2D-slices of the 3D mesh when oxygen availability takes value 6. B) The growth rate (over G) and lactate secretion (over G) as a function of glutamine availability (indicated as the glutamine over glucose ratio QGR), when the O2GR takes value 3 (O2:6 (mM h-1), G:3 (mM h-1)). C) Lactate efflux (z-axis) and PDH flux (color scale) scaled on glucose uptake as a function of QGR and O2GR. Red points highlight the O2GR level for which maximum lactate secretion is observed. Left inset: level of oxygen to be considered critical as a function of QGR. Critical oxygen is computed both as the point in which lactate secretion is at the maximum (blue dots) and the point in which PDH reaches its minimum (red dots). Right inset: pyruvate dehydrogenase flux as a function of glutamine availability (QGR) for different oxygen availability values (the O2GR value is reported on top of the corresponding curve).
Fig 5.
Flux patterns at critical O2GR.
A) ENGRO1 flux distribution that maximizes growth at critical O2GR (O2: 38 mM h-1; G: 10 mM h-1; Q: 40 mM h-1) as per S1 Table. The color and thickness of arrows scale with the flux intensity. Dashed grey lines are associated with null fluxes. The direction of the flux is indicated by the point of the arrow that is colored. Shaded shapes encompass reactions that show variability in optimal solutions. B-E) Examples of alternative flux patterns in optimal solutions.
Fig 6.
Qualitative representation of the TCA cycle flux mode associated with different optimization problems.
Red arrows indicate carbon fluxes deriving from glutamine, while blue arrows indicate carbon fluxes deriving from glucose. For each optimization, the value of the objective function value is reported on top of the scheme and it is normalized over the corresponding objective value of the control model. Left panels refer to the solution of biomass maximization problems, whereas right panels refer to the solution of ATP maximization problems. A-B) Optimal solutions of control model (O2: 38 mM h-1; G: 10 mM h-1; Q: 40 mM h-1). C-D) Optimal solutions when lactate secretion is inhibited. E-F) Optimal solutions when no glucose is available. G-H) Optimal solutions when no glucose is available, lactate production is blocked and pyruvate is allowed to accumulate.