Optimal cellular strategies for growth in uncertain environments

The study of cellular metabolism is of primary importance to understand molecular mechanisms governing cellular activities. Biologists made (and are still making) a big effort in characterizing numerous metabolic paths, nevertheless experimental results on cellular populations seem to be somehow in contradiction with their models. It is well established that glucose plays a central role in cellular foraging plans and it is known that mammalian cells can digest it mainly through two principal paths, so why don’t they always use the most efficient one among them (where efficiency is meant in terms of ATP yield)? This effect consisting in the choice of the "inefficient" metabolic pathway is known in literature as overflow metabolism and its understanding has immediate practical implications: cancer cells (as well as other fast reproducing cells) manifest it, so its understanding can lead to new diagnostic and therapeutic techniques for cancer. Since this "inefficient" phenotype is shared by almost all fast reproducing cells, its causes must be rooted in some general, fundamental constraint common to all cells. Vazquez et al. suggested that in every single cell, the finite solvent capacity of the cytoplasm can be the cause of this overflow metabolism; inspired by their work, we tried to generalize the model to populations of cells, making it more appropriate to describe a real-world situation. We assumed the cell to be a system detecting information from the environment (i.e. the glucose concentration) and elaborating an optimal response (i.e. the metabolic plan) while fulfilling the constraints imposed by the model itself and by rate-distortion theory for information transmission through channels. The proposed model leads to the profitable introduction of a key free parameter controlling phenotypic fluctuations. Predictions of the model are analyzed by conducting a careful numerical inspection of several configurations and results are in qualitative agreement with experiments. Eventually, making an analogy with free energy, we suggested that the model itself implicitly ensures the existence of an optimal parameter fine-tuned by evolution.