Evaluating the Potential of Applying Machine Learning Tools to Metabolic Pathway Optimization

Successful engineering of a microbial host for efficient production of a target product from a given substrate can be viewed as an extensive optimization task. Such a task involves the selection of high activity enzymes as well as their gene expression regulatory control elements (i.e., promoters and ribosome binding sites). Finally, there is also the need to tune expression of multiple genes along a heterologous pathway to relieve constraints from rate-limiting step and help reduce metabolic burden on cells from unnecessary over-expression of high activity enzymes. While the aforementioned tasks could be performed through combinatorial experiments, such an approach incurs significant cost, time and effort, which is a handicap that can be relieved by application of modern machine learning tools. Such tools could attempt to predict high activity enzymes from sequence, but they are currently most usefully applied in classifying strong promoters from weaker ones as well as combinatorial tuning of expression of multiple genes. This perspective reviews the application of machine learning tools to aid metabolic pathway optimization through identifying challenges in metabolic engineering that could be overcome with the help of machine learning tools.

Definitive screening for accelerated Taxol biosynthetic pathway optimization and scale up in Saccharomyces cerevisiae cell factories

Recent technological advancements in synthetic and systems biology have enabled the construction of microbial cell factories expressing diverse heterologous pathways in unprecedentedly short time scales. However, the translation of such laboratory scale breakthroughs to industrial bioprocesses remains a major bottleneck. In this study, an accelerated bioprocess development approach was employed to optimize the biosynthetic pathway of the blockbuster chemotherapy drug, Taxol. Statistical design of experiments approaches were coupled with an industrially relevant high-throughput microbioreactor system to optimize production of key Taxol intermediates, Taxadien-5α-ol and Taxadien-5α-yl-acetate, in engineered yeast cell factories. The optimal factor combination was determined via data driven statistical modelling and validated in 1L bioreactors leading to a 2.1-fold improvement in taxane production compared to a typical defined media. Elucidation and mitigation of a nutrient limitation enhanced product titers a further two-fold and titers of the critical Taxol precursors, Taxadien-5α-ol and Taxadien-5α-yl-acetate were improved to 34 and 11 mg/L, representing a three-fold improvement compared to the highest literature titers in S. cerevisiae. Comparable titers were obtained when the process was scaled up a further five-fold using 5 L bioreactors. The results of this study highlight the benefits of a holistic design of experiments guided approach to expedite early stage bioprocess development.

A Favorable Path to Domain Separation in the Orange Carotenoid Protein

The Orange Carotenoid Protein (OCP) is responsible for nonphotochemical quenching (NPQ) in cyanobacteria, a defense mechanism against potentially damaging effects of excess light conditions. This soluble two-domain protein undergoes profound conformational changes upon photoactivation, involving translocation of the ketocarotenoid inside the cavity followed by domain separation. Domain separation is a critical step in the photocycle of OCP because it exposes the N-terminal domain (NTD) to perform quenching of the phycobilisomes. Many details regarding the mechanism and energetics of OCP domain separation remain unknown. In this work, we apply metadynamics to elucidate the protein rearrangements that lead to the active, domain-separated, form of OCP. We find that translocation of the ketocarotenoid canthaxanthin has a profound effect on the energetic landscape and that domain separation only becomes favorable following translocation. We further explore, characterize, and validate the free energy surface (FES) using equilibrium simulations initiated from different states on the FES. Through pathway optimization methods, we characterize the most probable path to domain separation and reveal the barriers along that pathway. We find that the free energy barriers are relatively small (<5 kcal/mol), but the overall estimated kinetic rate is consistent with experimental measurements (>1 ms). Overall, our results provide detailed information on the requirement for canthaxanthin translocation to precede domain separation and an energetically feasible pathway to dissociation.

Engineering yeast for de-novo synthesis of the insect repellent - nepetalactone

ABSTRACTWhile nepetalactone, the active ingredient in catnip, is a potent insect repellent, its low in planta accumulation limits its commercial viability as an alternative repellent. Here we describe a platform for de novo nepetalactone production in Saccharomyces cerevisiae, enabling sustainable and scalable production. Nepetalactone production required introduction of eight exogenous genes including the cytochrome P450 geraniol-8-hydroxylase, which represented the bottleneck of the heterologous pathway. Combinatorial assessment of geraniol-8-hydroxylase and cytochrome P450 reductase variants, as well as copy-number variations were used to overcome this bottleneck. We found that several reductases improved hydroxylation activity, with a higher geraniol-8-hydroxylase ratio further increasing 8-hydroxygeraniol titers. Another roadblock was the accumulation of an unwanted metabolite that implied inefficient channeling of carbon through the pathway. With the native yeast old yellow enzymes previously shown to use monoterpene intermediates as substrates, both homologs were deleted. These deletions increased 8-hydroxygeraniol yield, resulting in a final de novo accumulation of 3.10 mg/L/OD600 of nepetalactone from simple sugar in microtiter plates. Our pathway optimization will aid in the development of high yielding monoterpene S. cerevisiae strains.

Biosensor-based multi-gene pathway optimization for enhancing the production of glycolate

Glycolate is widely used in industry, especially in the fields of chemical cleaning, cosmetics, and medical materials, and has broad market prospects for the future. Recent advances in metabolic engineering and synthetic biology have significantly improved the titer and yield of glycolate. However, an expensive inducer was used in previous studies that is not feasible for use in large-scale industrial fermentations. To constitutively biosynthesize glycolate, the expression level of each gene of the glycolate synthetic pathway needs to be systemically optimized. The main challenge of multi-gene pathway optimization is being able to select or screen the optimum strain from the randomly assembled library by an efficient high-throughput method within a short period of time. To overcome these challenges, we firstly established a glycolate-responsive biosensor and developed agar plate- and 48-well deep well plate-scale high-throughput screening methods for rapid screening of superior glycolate producers from a large library. A total of 22 gradient strength promoter-5′-UTR complexes were randomly cloned upstream of the genes of the glycolate synthetic pathway, generating a large random assembled library. After rounds of screening, the optimum strain was obtained from 6×105 transformants in a week, and it achieved a titer of 40.9 ± 3.7 g/L glycolate in a 5-L bioreactor. Furthermore, high expression levels of the enzymes YcdW and GltA were found to promote glycolate production, whereas AceA has no obvious impact on glycolate production. Overall, the glycolate biosensor-based pathway optimization strategy presented in this work provides a paradigm for other multi-gene pathway optimizations. Importance The use of strong promoters, such as pTrc and T7, to control gene expression not only need adding expensive inducers but also results in excessive protein expression that may be resulting in unbalanced metabolic flux and the waste of cellular building blocks and energy. To balance the metabolic flux of glycolate biosynthesis, the expression level of each gene needs to be systemically optimized in a constitutive manner. However, the lack of a high-throughput screening methods restricted the glycolate synthetic pathway optimization. Our work firstly established a glycolate-response biosensor, then agar plate and 48-well plate scale high-throughput screening methods were developed for rapid screening of optimum pathways from a large library. Finally, we obtained a glycolate producing strain with good biosynthetic performance, and the use of the expensive inducer IPTG was avoided, which broadens our understanding about the mechanism of glycolate synthesis.

Cell-based high-throughput screening of polysaccharide biosynthesis hosts

Valuable polysaccharides are usually produced using wild-type or metabolically-engineered host microbial strains through fermentation. These hosts act as cell factories that convert carbohydrates, such as monosaccharides or starch, into bioactive polysaccharides. It is desirable to develop effective in vivo high-throughput approaches to screen cells that display high-level synthesis of the desired polysaccharides. Uses of single or dual fluorophore labeling, fluorescence quenching, or biosensors are effective strategies for cell sorting of a library that can be applied during the domestication of industrial engineered strains and metabolic pathway optimization of polysaccharide synthesis in engineered cells. Meanwhile, high-throughput screening strategies using each individual whole cell as a sorting section are playing growing roles in the discovery and directed evolution of enzymes involved in polysaccharide biosynthesis, such as glycosyltransferases. These enzymes and their mutants are in high demand as tool catalysts for synthesis of saccharides in vitro and in vivo. This review provides an introduction to the methodologies of using cell-based high-throughput screening for desired polysaccharide-biosynthesizing cells, followed by a brief discussion of potential applications of these approaches in glycoengineering.

Increasing Solvent Tolerance to Improve Microbial Production of Alcohols, Terpenoids and Aromatics

Fuels and polymer precursors are widely used in daily life and in many industrial processes. Although these compounds are mainly derived from petrol, bacteria and yeast can produce them in an environment-friendly way. However, these molecules exhibit toxic solvent properties and reduce cell viability of the microbial producer which inevitably impedes high product titers. Hence, studying how product accumulation affects microbes and understanding how microbial adaptive responses counteract these harmful defects helps to maximize yields. Here, we specifically focus on the mode of toxicity of industry-relevant alcohols, terpenoids and aromatics and the associated stress-response mechanisms, encountered in several relevant bacterial and yeast producers. In practice, integrating heterologous defense mechanisms, overexpressing native stress responses or triggering multiple protection pathways by modifying the transcription machinery or small RNAs (sRNAs) are suitable strategies to improve solvent tolerance. Therefore, tolerance engineering, in combination with metabolic pathway optimization, shows high potential in developing superior microbial producers.