NEW CONSTRAINT-BASED APPROACHES TO TACKLE THE MULTIPLE SIDES OF CELL METABOLIC PLASTICITY AND HETEROGENEITY

Plasticity, heterogeneity and modelling approach constitute the three pillars on the top of which this thesis investigates the complexity of cell metabolism. The multiple sides of metabolic plasticity have been explored as cell adaptive response to varying conditions, demand and perturbations under both physiological and pathological conditions. By investigating cell populations as homogeneous and heterogeneous systems, new in silico predictive models and novel computational constraint-based methodologies have been defined. This work started from the investigation of cell populations as homogeneous systems, where the average behaviour is described and cell-to-cell differences are temporarily hidden. Reconstructing high-quality genome-scale metabolic models is crucial to computationally address cell metabolism and organize all the available metabolic knowledge of given cells or organisms. Although multiple tools for performing this task already exist, a pipeline for the semi-automatic reconstruction of genome-scale networks has been proposed to solve some current critical issues and generate higher quality models. The application of this approach for the genome-wide metabolic reconstruction of yeast Zygosaccharomyces parabailii showed adherence of in silico simulations to experimental data and literature findings. Moreover, metabolic plasticity in response to different metabolic regimes has been explored through constraint-based modelling. The potentialities of genome-scale reconstructions in mirroring the systemic perspective coexist with difficulty in their management. In this work, greater control is achieved by switching to smaller-scale core networks. In particular, core modelling has been exploited as an effective mean to investigate intertumoural heterogeneity, and plasticity of the implemented tumour metabolic programs as adaptation to different environmental scenarios. The effectiveness of homogeneous systems to lower overall system complexity level without compromising biological validity of in silico outcomes goes along with the need to address cell-to-cell variations of cell populations. In this regard, classic constraint-based modelling has been extended to deal with heterogeneous systems. A new strategy, called popFBA, has been developed to reconstruct and simulate cell populations metabolism, by putting emphasis on the relationships established among their components. Using as case study the ecosystemic view of cancer populations, popFBA highlighted that the achievement of optimal biomass is consistent with metabolic plasticity of population components under different scenarios together with a cooperative behaviour. At the same time, countless combinations of flux distributions for the individual population components prompted to develop a novel methodology called single-cell Flux Balance Analysis (scFBA). This metodology integrates single-cell transcriptomics data as further constraints on the individual components through the computation for each reaction of a Reaction Activity Score, which we implemented in a previous computational framework called MaREA. In this way, scFBA efficiently reduced the amount of allowable individual flux distributions, and captured complex networks of interactions between cells of a specific population. In view of the findings of this research, a deep characterization of metabolic plasticity within cell populations and of the intricate dialogue between cells and their environment can assist the formulation of more rational and personalized strategies. Their devising could enable to hamper disease progression, or to exploit metabolism of given microorganisms for producing relevant chemical compounds.