SSDesign: Computational metabolic pathway design based on flux variability using elementary flux modes

BayesianSSA: a Bayesian statistical model based on structural sensitivity analysis for predicting responses to enzyme perturbations in metabolic networks

BACKGROUND

Chemical bioproduction has attracted attention as a key technology in a decarbonized society. In computational design for chemical bioproduction, it is necessary to predict changes in metabolic fluxes when up-/down-regulating enzymatic reactions, that is, responses of the system to enzyme perturbations. Structural sensitivity analysis (SSA) was previously developed as a method to predict qualitative responses to enzyme perturbations on the basis of the structural information of the reaction network. However, the network structural information can sometimes be insufficient to predict qualitative responses unambiguously, which is a practical issue in bioproduction applications. To address this, in this study, we propose BayesianSSA, a Bayesian statistical model based on SSA. BayesianSSA extracts environmental information from perturbation datasets collected in environments of interest and integrates it into SSA predictions.

RESULTS

We applied BayesianSSA to synthetic and real datasets of the central metabolic pathway of Escherichia coli. Our result demonstrates that BayesianSSA can successfully integrate environmental information extracted from perturbation data into SSA predictions. In addition, the posterior distribution estimated by BayesianSSA can be associated with the known pathway reported to enhance succinate export flux in previous studies.

CONCLUSIONS

We believe that BayesianSSA will accelerate the chemical bioproduction process and contribute to advancements in the field.

Recent advances in metabolic engineering-integration of in silico design and experimental analysis of metabolic pathways

Microorganisms are widely used to produce valuable compounds. Because thousands of metabolic reactions occur simultaneously and many metabolic reactions are related to target production and cell growth, the development of a rational design method for metabolic pathway modification to optimize target production is needed. In this paper, recent advances in metabolic engineering are reviewed, specifically considering computational pathway modification design and experimental evaluation of metabolic fluxes by ¹³C-metabolic flux analysis. Computational tools for seeking effective gene deletion targets and recruiting heterologous genes are described in flux balance analysis approaches. A kinetic model and adaptive laboratory evolution were applied to identify and eliminate the rate-limiting step in metabolic pathways. Data science-based approaches for process monitoring and control are described to maximize the performance of engineered cells in bioreactors.

Dynamic multiscale metabolic network modeling of Chinese hamster ovary cell metabolism integrating N-linked glycosylation in industrial biopharmaceutical manufacturing

Experimental and modeling work, described in this article, is focused on the metabolic pathway of Chinese hamster ovary (CHO) cells, which are the preferred expression system for monoclonal antibody protein production. CHO cells are one of the primary hosts for monoclonal antibodies production, which have extensive applications in multiple fields like biochemistry, biology and medicine. Here, an approach to explain cellular metabolism with in silico modeling of a microkinetic reaction network is presented and validated with unique experimental results. Experimental data of 25 different fed-batch bioprocesses included the variation of multiple process parameters, such as pH, agitation speed, oxygen and CO₂ content, and dissolved oxygen. A total of 151 metabolites were involved in our proposed metabolic network, which consisted of 132 chemical reactions that describe the reaction pathways, and include 25 reactions describing N-glycosylation and additional reactions for the accumulation of the produced glycoforms. Additional eight reactions are considered for accumulation of the N-glycosylation products in the extracellular environment and one reaction to correlate cell degradation. The following pathways were considered: glycolysis, pentose phosphate pathway, nucleotide synthesis, tricarboxylic acid cycle, lipid synthesis, protein synthesis, biomass production, anaplerotic reactions, and membrane transport. With the applied modeling procedure, different operational scenarios and fed-batch techniques can be tested.

Application of theoretical methods to increase succinate production in engineered strains

Computational methods have enabled the discovery of non-intuitive strategies to enhance the production of a variety of target molecules. In the case of succinate production, reviews covering the topic have not yet analyzed the impact and future potential that such methods may have. In this work, we review the application of computational methods to the production of succinic acid. We found that while a total of 26 theoretical studies were published between 2002 and 2016, only 10 studies reported the successful experimental implementation of any kind of theoretical knowledge. None of the experimental studies reported an exact application of the computational predictions. However, the combination of computational analysis with complementary strategies, such as directed evolution and comparative genome analysis, serves as a proof of concept and demonstrates that successful metabolic engineering can be guided by rational computational methods.

Computing the various pathways of penicillin synthesis and their molar yields

More than 80 years after its discovery, penicillin is still a widely used and commercially highly important antibiotic. Here, we analyse the metabolic network of penicillin synthesis in Penicillium chrysogenum based on the concept of elementary flux modes. In particular, we consider the synthesis of the invariant molecular core of the various subtypes of penicillin and the two major ways of incorporating sulfur: transsulfuration and direct sulfhydrylation. 66 elementary modes producing this invariant core are obtained. These show four different yields with respect to glucose, notably ½, 2/5, 1/3, and 2/7, with the highest yield of ½ occurring only when direct sulfhydrylation is used and α-aminoadipate is completely recycled. In the case of no recycling of this intermediate, we find the maximum yield to be 2/7. We compare these values with earlier literature values. Our analysis provides a systematic overview of the redundancy in penicillin synthesis and a detailed insight into the corresponding routes. Moreover, we derive suggestions for potential knockouts that could increase the average yield.

Co-evolution of strain design methods based on flux balance and elementary mode analysis

More than a decade ago, the first genome-scale metabolic models for two of the most relevant microbes for biotechnology applications, Escherichia coli and Saccaromyces cerevisiae, were published. Shortly after followed the publication of OptKnock, the first strain design method using bilevel optimization to couple cellular growth with the production of a target product. This initiated the development of a family of strain design methods based on the concept of flux balance analysis. Another family of strain design methods, based on the concept of elementary mode analysis, has also been growing. Although the computation of elementary modes is hindered by computational complexity, recent breakthroughs have allowed applying elementary mode analysis at the genome scale. Here we review and compare strain design methods and look back at the last 10 years of in silico strain design with constraint-based models. We highlight some features of the different approaches and discuss the utilization of these methods in successful in vivo metabolic engineering applications.

•

Computational strain design methods are divided into two main families.

•

We trace the evolutionary history of these two families.

•

Surveyed successful cases of model-guided strain design for industrial applications.

•

Most proposed methods have not yet been tested in real applications.

•

Agreement between in silico and in vivo results shows potential of tested methods.