Fig 1.
Metabolic symbiosis in tumour.
Schematic diagram and flow chart illustrating the metabolic symbiosis mechanism between oxygenated and hypoxic tumour cells. Oxygenated cells at the tumour boundary (shown as green cells) consume exogenous lactate via MCT1 transporters and undergo lactate metabolism through OXPHOS. Inner hypoxic cells (shown as brown cells) consume glucose through GLUT transporters and undergo glycolysis and release lactate into the tumour microenvironment using MCT4 transporters. Metabolic symbiosis between the two cell populations (green cells and brown cells) helps hypoxic cells increase their glucose uptake, thereby helping tumour cells survive under low glucose conditions (created with BioRender.com).
Fig 2.
Agent-based mathematical model.
(A). The model is a multi-scale agent-based model implemented using the NetLogo platform. The extra-cellular scale models gradients of substances such as oxygen, glucose and lactate by using partial differential equations. The cellular scale models cell-cell interactions using a cellular automaton approach. The smallest scale, the intra-cellular scale, handles subcellular molecular interactions using a Boolean gene regulatory network. All scales communicate with each other and therefore the tumour growth is an emerging property of the sub-cellular molecular interactions. The model can be used to study tumour growth under different environmental conditions, gene alterations and heterogeneous cell populations (created with BioRender.com). (B). The model flow chart shows how the three different scales are connected. If the local oxygen and glucose levels are below their respective threshold values, the cell becomes necrotic. If the cell is not in the necrotic state, the intra-cellular regulatory network determines the cell fate which is Proliferation or Apoptosis or Growth Arrest. An apoptotic cell is removed from the simulation immediately. A proliferative cell can divide if there is empty space nearby, otherwise it can switch to growth arrest state and waits a TQ time before checking environmental conditions again to find its new phenotype. The cell phenotype can influence the gradients of the diffusible substances on the microenvironment through producing/consuming diffusible substances based on its phenotype. The altered environmental properties are fed to the intra-cellular network of the cell again through input nodes and then the network can decide its new phenotype according to altered microenvironmental conditions.
Fig 3.
(A). If MCT1 is not mutated (MCT1wt), symbiosis is observed, while when MCT1 is mutated with loss of function (MCT1-), symbiosis is not observed. Different cell populations are shown: OXPHOS (blue cells, glucose/lactate—> pyruvate—>OXPHOS), Glycolysis (green cells, glucose—>pyruvate—>lactate), both OXPHOS and glycolysis (purple cells), not metabolically active or quiescence (gray cells). Gradients of glucose, lactate, tumour growth factor alpha (TGFA) and pH are shown on the microenvironment. (B, C). Number of total, glycoATP and mitoATP cells are shown. The total number of cells contains all the cells in the tumour including dead cells. The mitoATP and glycoATP are the number of cells that rely on OXPHOS and glycolysis for ATP production, respectively. (D, E). Number of hypoxic and oxygenated tumour cells with MCT1 wild type and mutated condition are shown. The hypoxic cell population is not seen when symbiosis is lost (MCT1-) because oxygen cannot be depleted fast enough without lactate oxidation. (F, G). Metabolic symbiosis index and number of active (viable) cells are shown when tumour grows with MCT1wt and MCT1-, so with and without symbiosis. The metabolic symbiosis index quantifies the strength of symbiosis between hypoxic and oxygenated tumour cells. The index can range from 0 to 1 depending on the symbiosis strength. The active cells are the cells with their metabolic pathways are active (i.e., blue, green and purple cells shown in A). The shaded area of curves shows respective standard deviation.
Fig 4.
Metabolic symbiosis simulations with network gene alterations: The gene enriched (+) and knockout (-) status were simulated by setting the respective node of the regulatory network to 1 and 0, respectively.
The gene wild type (WT) status was simulated without setting the respective node to either 0 or 1. Each gene was altered individually. (A). Percentage tumour growth increase due to symbiosis is shown for each gene enrichment status. (B). Percentage tumour growth increase due to symbiosis is shown for each gene knockout status. (C). Whether tumour growth is significantly different (p-value < 0.05) between symbiosis and non-symbiosis for each gene enrichment status is shown. (D). Whether tumour growth is significantly different (p-value < 0.05) between symbiosis and non-symbiosis for each gene knockout status is shown. Clusters of gene alterations can be identified, which enhance tumour growth due to symbiosis while some other gene alterations together with symbiosis adversely affect tumour growth (A, B). Colours indicate percentage growth increase by symbiosis (A, B) and p values (C, D). p values from 0 to 0.05 are shown in red to white colour scale and p values ≥ 0.05 are shown in grey colour.
Fig 5.
Tumour growth under wild type and mutated p53 status.
Effect of p53 status on OXPHOS (shown as mitoATP), glycolysis (shown as glycoATP) and ATP production rate are shown. p53wt and p53- cells were grown in isolation. The starting number of cells was 100. (A, B). Growth of p53wt (A) and p53- (B) cells over time. The OXPHOS cell population is marginally dominant over glycolytic population regardless of the p53 status. (C). Evolution of OXPHOS/Glycolysis cell ratio with p53wt and p53- status. (D, E). Variation of OXPHOS (mitoATP) and glycolytic (mitoATP) ATP production rate of p53wt (D) and p53- (E) cells as tumour grows. The tumour cells with p53 knockout were found to grow faster than p53wt cells because they lacked p53 to suppress tumour growth.
Fig 6.
Distribution of tumour microenvironmental substances under wild type and mutated p53 status.
(A-D). Heat maps show spatial-temporal variation of microenvironmental (A) oxygen, (B) glucose, (C) lactate, and (D) pH levels for p53wt and p53- tumour growth (here, the Length is the cross section of the tumour microenvironment through the center of the tumour). Red lines show the development of tumour boundary over time. The results show that p53 loss of function could make the tumour microenvironment more acidic during the period of time (i.e., 25 days) considered here. This is mainly because the p53- tumour grows faster and therefore more waste products are accumulated inside the tumour making it more acidic.
Fig 7.
Symbiosis-induced percentage change of glucose and oxygen in the tumour microenvironment.
(A-B). Percentage change of (A) glucose and (B) oxygen in the tumour microenvironment over time for tumour growth with p53wt and p53- status. Length and red lines are as described in Fig 6. The model was run with MCT1wt and MCT1- status. The percentage change of glucose and oxygen in the tumour microenvironment was calculated over time. The heat maps suggest that symbiosis increases glucose level while decreasing oxygen level at the tumour boundary. More supplementary results are shown in S8 Fig.
Fig 8.
Competition between symbiotic and non-symbiotic tumour cells.
Graphs depicting the number of active cells for two p53- cell populations with or without symbiotic capacity, with MCT1wt and MCT1-, respectively. The ratio of MCT1wt: MCT1- was 1:1 at the start of the simulations. Cells were allowed to compete at 6% of oxygen and varying levels of glucose and lactate maintained at the boundary of the computational domain. The number of total, active MCT1wt and active MCT1- cells are shown. (A). 1mM glucose and 1 mM lactate, (B). 1mM glucose and 5 mM lactate, (C). 5 mM glucose and 1 mM lactate, (D). 5 mM glucose and 5 mM lactate. The results show that symbiotic cells would not get a significant competitive advantage over non-symbiotic cells if there is sufficient glucose in the tumour microenvironment.
Fig 9.
MCT1 and GLUT1 inhibition with continuous drug administration.
Heatmaps depicting how MCT1 and GLUT1 inhibitors (denoted as MCT1i and GLUT1i) interfere with glycolysis, OXPHOS, and tumour growth. The MCT1i concentration, [MCT1i], was varied up to 1000 times of its half-maximal inhibitory concentration (IC50) value while GLUT1i concentration, [GLUT1i], was varied up to 100 times of its IC50 value. Inhibitor concentration was maintained at the boundary of the computational domain throughout the simulation time. (A). Number of total cells, OXPHOS cell population and glycolytic cell population of p53wt tumour after 25 days of growth. (B). Number of total cells, OXPHOS cell population and glycolytic cell population of p53- tumour after 25 days of growth. The initial tumour consisted of 100 cells, with either with p53wt or p53- status.
Fig 10.
MCT1 and GLUT1 inhibition with alternative and periodic drug administration.
Heatmaps depicting how MCT1 and GLUT1 inhibitors interfere with glycolysis, OXPHOS, and tumour growth when both drugs were given alternatively and periodically. Both MCT1i and GLUT1i concentrations were kept at same level, and it was maintained at 1, 10 and 100 of respective half-maximal inhibitory concentration (IC50) value. Inhibitor concentration was maintained at the boundary of the computational domain and varied periodically throughout the simulation time. The initial tumour consisted of 100 cells, with p53- status. Number of total cells, OXPHOS cell population and glycolytic cell population of p53- tumour after 30 days of growth are shown. The period “NA” represents the simulation with continuous supply of both drugs at half of the respective concentration.
Fig 11.
MCT1 and GLUT1 inhibition with simultaneous and periodic drug administration.
Heatmaps depicting how MCT1 and GLUT1 inhibitors interfere with glycolysis, OXPHOS, and tumour growth when both drugs were given simultaneously and periodically. Other simulation conditions remained consistent with those in Fig 10. Number of total cells, OXPHOS cell population and glycolytic cell population of p53- tumour after 30 days of growth are shown. The period “NA” represents the simulation with continuous supply of both drugs at half of the respective concentration.