Metabolic Model Based Computational Systems Biological Research For Single And Mixed Microbes

Building mathematical model and developing associated algorithms are of great significance to researchers in understanding and utilizing living cells.Modelling and computational researches for living cells serve not only the purpose of modifying microbe into efficient ’chemical factory’,but also the purpose of gaining a deeper understanding in the mechanism of living machines.Since living cells have very sophisticated structure,the relatively simple model microbes with solid research background become a good starting point.Since the beginning of genome era,genome scale metabolic models and associated algorithms for systematic analysis have been developed rapidly;new methods keep emerging to develop models towards describing cells’reality as close as possible,and applications of metabolic models keeps appearing in publications.However there are many fundamental problems remain to be solved in this field,for example,many metabolic models need improment in quality,but there is no efficient and systematic method to correct the models;many literatures address simulation of cell growth,but seldom focus on predicting mortality rate of cell;microbial metabolic interactions are ubiquitous in nature,but current methods to predict it has major problem,leading to significant prediction error.This paper will address the aforementioned problems by updating/merging theories,building new models,and inventing new algorithms.This paper consists of two major parts,the first part(chapter two,for ’single species’)will be dedicated to corrections of available metabolic models for single species,systematically evaluating different heterogonous pathways for overproduction of specific chemicals and developing software for flux simulation and metabolic engineering strategies predictions.The second part(chapter three,the ’multi-species part’)will dedicate to developing novel algorithm for predicting microbial metabolic interactions and limiting crossfeeding to community member’s growthIn the‘single species’part,a novel algorithm called biomassMW will be introduced and used to systematically correct errors exist in a previous version of the genome scale metabolic models of C.glutamicum and C.Reinhardtii.The corrections including correcting large amount of reactions with mass and/or charge imbalance,computing missing metabolites formula,correcting biomass equations and computing and standardizing biomass formula,new versions of metabolic model of these two species were obtained.Comparison of flux prediction by the new C.glutamicum model CG3 and 13C experimental fluxes indicates CG3 achieve 0.43 mmol/gDWh as the root mean squared error(RSME)in predicting intracellular fluxes,99.3%lower than the RSME of the last version published in 2009.After confirming that CG3 successfully predicted the metabolic engineering strategies for overproduction of lysine reported by literatures,the model was used to design strategies for overproduction of proline including relieving fed-back inhibition,down-regulation of competitive pathways and enhancing precursor supplies,the strategies were successfully lead to development of a successful strain by our collaborator,achieving 20g/L of flask fermentation concentration and 50g/L(7.5L,60h)of fermentor concentration.Then we applied the model to pathway evaluation for overproduction of adipic acid,the process includes evaluating thermodynamical feasibility,comparing theoretical yield and supplying-demanding balance of reductive co-factors,the optimal pathway is then selected because it can utilize reductive co-factor more efficiently under optimal conditions and reducing waste of carbon flux than other options.Further predicted aims to find metabolic intervening strategies for increasing adaptation of the host cell and selected heterologous pathways,we integrated the methods aforementioned into a module-based software for convient use.Building upon the models developed in the part one,part two is dedicated to studying microbial metabolic interactions,which are ubiquitous in the world of microbes.However,metabolite exchanges,a major type of microbial interactions,remain difficult to measure and predict,invoking the urgent need of modeling and computational studies.As an alternative to the conventional ecological models,which usually lack metabolic details,metabolic models emerge as a promising way to address the challenge.However,existing algorithms for predicting microbial community metabolism usually impose constraints or objective functions(implicitly or explicitly)that lead to ’forced altruism’,which forces a microbe to fulfill other species’ need by exporting certain metabolites before its own in order to computationally achieve optimization of community level objective,as a result,from the view of game theory,the prediction is not necessarily Nash equilibrium.We developed a bi-level optimization framework free of ’forced altruism’constraints termed NEcom,which is proved to guarantee prediction of Nash equilibria of microbial interactions.The inner level of NEcom optimizes the individual fitness while the outer level is designed to find Nash equilibria.Meanwhile a complementary shadow-price-based iterative algorithm(FShaP)was developed to predict state transitions and draw analogy to traditional payoff matrix analysis,bridging the gap between traditional game theory and metabolic models.By comparing with the results obtained by FShaP,we demonstrated that the evolutionary steady states(ESS)of several classical games can be predicted by NEcom,including prisoner’s dilemma,snowdrift game and positive frequencydependent cooperation.A reported algae-yeast co-culture system has been analyzed using NECom.More than 1200 growth conditions were simulated,of which 488 conditions correspond to 3221 experimental data points with an overall 63.5%and 81.7%reduction in root-mean-square error for the two species respectively when compared with the standard flux balance analysis,showing that NECom can capture and explain experimental trends using a minimal amount of experimental data as inputs without ad-hoc parameters.NECom is able to reveal that the decreasing glucose concentration vs.the constant photon availability provides significant growth advantage of C.reinhardtii over S.cerevisiae;CO2 and ammonia are growth-limiting cross-feeding substrates for C.reinhardtii and S.cerevisiae respectively;and the negative frequency-dependent growth pattern is caused by opposite trends of the supply rates of each species’ limiting crossfeedings.The results demonstrate the power of NEcom to predict real world microbial interactions and shed new light to revealing the metabolic mechanisms of microbial interactions.