Abstract
Engineering Microbial Cell Factories
Building upon recent successes in bioprocessing that use novel microbial strains to produce value-added chemicals, industrial biotechnology is poised to benefit further from computational models that can inform the design of microbes as enhanced cell factories. Prominent successes in the application of industrial biotechnology include the long-term use of Chinese hamster ovary (CHO) cells for the production of recombinant therapeutic proteins, 1 engineered yeast and Escherichia coli for the production of the malarial drug precursor artemisinin, 2 the development of modified strains of E. coli and Clostridia for the (enhanced) production of butanol, 3,4 and the use of engineered algae strains for biofuel production. 5 In each of these cases, active research efforts seek to apply computational models to improve the rational design of cell factories, and the research has progressed to the point at which companies have been founded and are progressing substantially towards bringing products to market.
The use of microbes as cell factories for the production of a variety of industrially important chemicals has risen in prominence over the past several years, and the growth of this industry is accelerating. For example, Genomatica (San Diego, CA) has developed a microorganism that can consume simple sugars and produce butanediol in a single process step. 6 This chemical is important to such industrial processes as the production of materials for athletic shoes and spandex. Amyris (Emeryville, CA) has developed microbial strains capable of producing artemisinic acid, an important precursor to the antimalarial drug artemisinin. 7 Amyris is currently scaling its process to achieve microbial production to meet the world's demand for atemisinic acid, thus enabling the low-cost production of anti-malarial medication. Such an achievement clearly has huge implications for global health and would be a major success for industrial biotechnology. Companies such as Sapphire Energy (San Diego, CA) are engineering algae to generate biofuels from CO2 at large scale, helping to generate sustainable and environmentally friendly energy. 8 The increasing number of such success stories in the field of industrial biotechnology has been fueled in part by the genomics revolution and the power of molecular and systems biology approaches. Modeling efforts have the potential to accelerate this growth, enabling the rational design of microbial cell factories that can be quickly engineered and rapidly scaled beyond the laboratory bench.
One systems approach that increasingly informs microbial designs is genome-scale computational modeling of cellular metabolism. Computational models provide a framework for interpreting data from experiments and can form the basis for rational design strategies for biotechnology. For example, Genomatica's successful strategy is built, in part, on constraint-based models of metabolism as a means of accelerating strain design. Many of the methods to reconstruct and analyze such models have been formalized, catalogued, and are distributed in several open-source environments. 9 –12 Over the past decade or so, the field of metabolic modeling has grown considerably, essentially moving through three phases of development: inception and definition; expansion of methods and building the theoretical basis; and applications to real-world problems. It is the ongoing work in this third area that brings genome-scale modeling increasingly into the realm of Industrial Biotechnology. In this special issue of Industrial Biotechnology, we examine this emerging field as a key driver for the development and commercial application of cell factories.
Introduction to Special Research Section
This Special Research Section comprises a set of reviews that provides updates on the progress of computational modeling of microbial factories, as well as research articles that present new metabolic modeling capabilities.
Smallbone and Mendes review classes of large-scale metabolic models, ranging from network reconstruction to analysis using constraint-based approaches and beyond towards fully parameterized kinetic models. Garcia-Albornoz and Nielsen discuss the application of genome-scale metabolic models in a variety of metabolic engineering contexts. Harrison and Herrgard review the present uses and future prospects of applying large-scale measurements of a cell's metabolome—one of the most important emerging data types for developing enhanced cellular factories. Finally, Koskimaki et al. discuss the application of such models to guide engineering efforts in the case of harnessing algae for industrial applications.
The research contributions in this special issue describe new optimization-based techniques for analyzing genome-scale metabolic models, report a new application of such models, and provide insights to the iterative process of model development and improvement. King and Feist present a new method, OptSwap, to help optimize cofactor specificity of oxidoreductase reactions for the generation of microbial production strains. Hyland et al. use a genome-scale model of the yeast Saccharomyces cerevisiae to predict weak acid toxicity. Finally, Aung et al. report the enhancement of the metabolic reconstruction for yeast by more comprehensively mapping fatty acid, glycerolipid, and glycerophospholipid metabolism.
As these reviews and research contributions demonstrate, the development of genome-scale metabolic models typically passes through distinct phases. First, in the network reconstruction phase, all of the known relevant information about the metabolism of the organism is assembled into a cohesive whole. Second, the computational model is curated, and missing capabilities are identified and filled until the model can simulate such properties as growth on different substrates or byproduct secretion rates. With this capability, it then becomes possible to compare model predictions to data, and this typically results in an iterative period of experimentation and prediction until a model of sufficient quality is achieved. Once the model is validated, it can then be used to guide experimental strain design. The models can simulate the effects of knockouts, gene additions, co-factor specificity changes and so forth, and thus provide a roadmap for modifications that can take substantial effort and time to implement. As model capabilities improve over time, the ability of the model to inform the rational design of microbial cell factories also improves. I hope that this special issue will provide you a useful glimpse into the current state and future promises of genome-scale metabolic modeling for industrial biotechnology.