Constraint-based modeling of metabolism - interpreting predictions of growth and ATP synthesis in human and yeast
Doctoral thesis, 2019

Growth is the primary objective of the cell. Diseases arise when cells diverge from a healthy growth-pattern. An increased understanding of cellular growth may thus be translated into improved human health. The cell requires materials and free energy (in the form of ATP) in order to grow, metabolism supplies the cell with this. The rate of metabolism is ultimately constrained by the biophysical properties of the metabolic enzymes. Interactions between the constraints and the growth-objective gives rise to metabolic trade-offs, e.g. between ATP synthesis from respiration and fermentation. We can gain quantitative insight into these processes by simulating metabolism using mathematical models. In this thesis I simulated the metabolism of four biological systems: the infant, cancer, yeast and muscle. The simulations demonstrated how a shift in metabolic strategy may increase the rates of ATP synthesis and growth. These increased metabolic rates come at the expense of decreased resource efficiency, i.e. ATP produced per carbon consumed. The effect was primarily caused by the low catalytic efficiency of the respiratory enzyme complexes I and V. By shifting from respiratory to fermentative ATP synthesis, the cell was able to bypass these constraints. An intermediate strategy involved bypassing only complex I. The phenomenon was experimentally corroborated in the working muscle, and it is the native state of the yeast Saccharomyces cerevisiae (which lacks complex I). The differences in efficiency between the different metabolic pathways also explained why cells grow faster on some carbon sources, e.g. the specific growth rate for yeast is higher on glucose than on ethanol. These models were extended to predict the world-record running-speeds at different distances, by taking the sizes of the body’s nutrient-deposits into account. A metabolic strategy employed by cancer cells involved excretion of the amino acid glutamate. The simulations showed a mechanistic relation to catabolism of branched-chain amino acids and the localization of amino acid metabolism to different cellular compartments. By experimentally inhibiting glutamate excretion using an off-the-shelf drug (sulfasalazine), the growth rate of a cancer cell line was reduced. The metabolic modeling involved integration of various types of data and thus demonstrated the potential to unify knowledge from different studies and domains. This exposed contradictory claims in literature and highlighted knowledge-gaps that need to be filled to further improve human health.

the Crabtree effect



cellular economy

the Warburg effect



flux balance analysis

lactate threshold

metabolite depletion

KC-salen, Kemihuset, Kemigården 4, Göteborg
Opponent: associate professor Tomer Shlomi, Department of Computer Science and Biology, Technion - Israel Institute of Technology, Israel.


Avlant Nilsson

Chalmers, Biology and Biological Engineering, Systems and Synthetic Biology

Predicting growth of the healthy infant using a genome scale metabolic model.

npj Systems Biology and Applications,; Vol. 3(2017)p. 3-

Journal article

Nilsson, A., Haanstra, J.R., Engqvist, M., Gerding, A., Bakker, B., Klingmüller, U., Teusink, B. and Nielsen, J. Liver cancer cells excrete glutamate of cytosolic but not mitochondrial origin

Metabolic Trade-offs in Yeast are Caused by F1F0-ATP synthase

Scientific Reports,; Vol. 6(2016)

Journal article

Nilsson, A., Björnson, E., Flockhart, M., Larsen, F., Nielsen, J. Complex I is bypassed during high intensity exercise

Metabolic Models of Protein Allocation Call for the Kinetome

Cell Systems,; Vol. 5(2017)p. 538-541

Other text in scientific journal

Every human develops from a single cell. The cell grows and divides billions of times. An imbalance in this process may affect health. One example is stunting. Children that are malnourished grow slower than healthy children. Another example is cancer. Cancer cells grow much faster than healthy cells. Cell growth depends on chemical processes known as metabolism. Metabolism transforms nutrients into the building blocks of the cell. The blocks are joined using energy delivered by the molecule ATP. ATP is charged with energy by metabolism.  The energy originates from the breakdown of nutrients. Metabolism is made up of many different enzymes. These are small proteins that catalyze chemical reactions in the cell. The metabolism of a diseased cell is often different from a healthy. Cancer cells consume nutrients faster than healthy cells, as a result they also release more byproducts. Lactate is one of the byproducts, which is also produced in muscles when they are working hard. Cancer cells also release the byproduct glutamate, but it is unknown why. By studying metabolism we can learn more about growth and disease. But human cells are challenging to study. One approach is to study simpler organisms. Baker’s yeast, Saccharomyces cerevisiae, is the source of much of our knowledge about how growing cells divide. Another approach is to model the cells in a computer. An example is constraint based models of metabolism. They are used to simulate how cells are affected by physical constraints, both internal and external. An internal constraint may be that that different types of enzymes work with different speed. An external constraint may be if a nutrient is available or not.

I used constraint based models of metabolism to study growth and production of ATP. Several findings came out of the simulations. 1) Breastfed infant’s growth is constrained by the rate of ATP production. 2) Cells can switch enzymes to increase the rate of ATP production. But this comes at the cost of a more wasteful extraction of nutrients. The behavior of both yeast cells, cancer cells and human muscles can be explained by this. 3) The rate of growth can be predicted for cells that are fed different nutrients. 4) Cancer cells excrete glutamate because it is a byproduct of growth. The rate of growth in cancer cells can be decreased by blocking glutamate excretion.

Computer models are powerful tools. They allow knowledge to be integrated from different scientific studies and experiments. The models make predictions about reality. By interpreting the predictions it is possible to extract knowledge that may have been overlooked. Computer models highlight gaps in human knowledge. They advance our understanding of cells towards future improvements of human health.

Integrating Modelling of Metabolism and Signalling towards an Application in Liver Cancer (ERASysAPP - IMOMESIC)

Swedish Research Council (VR) (2014-6544), 2014-10-01 -- 2018-09-30.

Subject Categories

Cell Biology

Biochemistry and Molecular Biology

Medical Biotechnology (with a focus on Cell Biology (including Stem Cell Biology), Molecular Biology, Microbiology, Biochemistry or Biopharmacy)

Bioinformatics and Systems Biology


Basic sciences

Areas of Advance

Life Science Engineering (2010-2018)



Doktorsavhandlingar vid Chalmers tekniska högskola. Ny serie: 4534



KC-salen, Kemihuset, Kemigården 4, Göteborg

Opponent: associate professor Tomer Shlomi, Department of Computer Science and Biology, Technion - Israel Institute of Technology, Israel.

More information

Latest update