Systems Metabolic Engineering Meets Machine Learning: A New Era for Data-Driven Metabolic Engineering

The recent increase in high-throughput capacity of 'omics datasets combined with advances and interest in machine learning (ML) have created great opportunities for systems metabolic engineering. In this regard, data-driven modeling methods have become increasingly valuable to metabolic strain design. In this review, the nature of 'omics is discussed and a broad introduction to the ML algorithms combining these datasets into predictive models of metabolism and metabolic rewiring is provided. Next, this review highlights recent work in the literature that utilizes such data-driven methods to inform various metabolic engineering efforts for different classes of application including product maximization, understanding and profiling phenotypes, de novo metabolic pathway design, and creation of robust system-scale models for biotechnology. Overall, this review aims to highlight the potential and promise of using ML algorithms with metabolic engineering and systems biology related datasets.

Design and synthesis of synthetic UP elements for modulation of gene expression in Escherichia coli

Metabolic engineering requires fine-tuned gene expression for most pathway optimization applications. To develop a suitable suite of promoters, traditional bacterial promoter engineering efforts have focused on modifications to the core region, especially the −10 and −35 regions, of native promoters. Here, we demonstrate an alternate, unexplored route of promoter engineering through randomization of the UP element of the promoter—a region that contacts the alpha subunit carboxy-terminal domain instead of the sigma subunit of the RNA polymerase holoenzyme. Through this work, we identify five novel UP element sequences through library-based searches in Escherichia coli. The resulting elements were used to activate the E. coli core promoter, rrnD promoter, to levels on par and higher than the prevalent strong bacterial promoter, OXB15. These relative levels of expression activation were transferrable when applied upstream of alternate core promoter sequences, including rrnA and rrnH. This work thus presents and validates a novel strategy for bacterial promoter engineering with transferability across varying core promoters and potential for transferability across bacterial species.

Metabolic engineering of microbial cell factories for production of nutraceuticals

Metabolic engineering allows for the rewiring of basic metabolism to overproduce both native and non-native metabolites. Among these biomolecules, nutraceuticals have received considerable interest due to their health-promoting or disease-preventing properties. Likewise, microbial engineering efforts to produce these value-added nutraceuticals overcome traditional limitations of low yield from extractions and complex chemical syntheses. This review covers current strategies of metabolic engineering employed for the production of a few key nutraceuticals with selecting polyunsaturated fatty acids, polyphenolic compounds, carotenoids and non-proteinogenic amino acids as exemplary molecules. We focus on the use of both mono-culture and co-culture strategies to produce these molecules of interest. In each of these cases, metabolic engineering efforts are enabling rapid production of these molecules.

CRISPR-PIN: Modifying gene position in the nucleus via dCas9-mediated tethering

Spatial organization of DNA within the nucleus is important for controlling DNA replication and repair, genetic recombination, and gene expression. Here, we present CRISPR-PIN, a CRISPR/dCas9-based tool that allows control of gene Position in the Nucleus for the yeast Saccharomyces cerevisiae. This approach utilizes a cohesin-dockerin interaction between dCas9 and a perinuclear protein. In doing so, we demonstrate that a single gRNA can enable programmable interaction of nuclear DNA with the nuclear periphery. We demonstrate the utility of this approach for two applications: the controlled segregation of an acentric plasmid and the re-localization of five endogenous loci. In both cases, we obtain results on par with prior reports using traditional, more cumbersome genetic systems. Thus, CRISPR-PIN offers the opportunity for future studies of chromosome biology and gene localization.

•

dCas9 artificially localized to nuclear periphery using cohesin-dockerin tether to ESC1.

•

Targeting dCas9 to acentric plasmid allows rescue of plasmid segregation phenotype.

•

5 unique chromosomal loci re-localized to nuclear periphery.

•

dCas9 tethering allows control over target gene Position In the Nucleus (CRISPR-PIN).

Ameliorating the Metabolic Burden of the Co-expression of Secreted Fungal Cellulases in a High Lipid-Accumulating Yarrowia lipolytica Strain by Medium C/N Ratio and a Chemical Chaperone

Yarrowia lipolytica, known to accumulate lipids intracellularly, lacks the cellulolytic enzymes needed to break down solid biomass directly. This study aimed to evaluate the potential metabolic burden of expressing core cellulolytic enzymes in an engineered high lipid-accumulating strain of Y. lipolytica. Three fungal cellulases, Talaromyces emersonii-Trichoderma reesei chimeric cellobiohydrolase I (chimeric-CBH I), T. reesei cellobiohydrolase II (CBH II), and T. reesei endoglucanase II (EG II) were expressed using three constitutive strong promoters as a single integrative expression block in a recently engineered lipid hyper-accumulating strain of Y. lipolytica (HA1). In yeast extract-peptone-dextrose (YPD) medium, the resulting cellulase co-expressing transformant YL165-1 had the chimeric-CBH I, CBH II, and EG II secretion titers being 26, 17, and 132 mg L^-1, respectively. Cellulase co-expression in YL165-1 in culture media with a moderate C/N ratio of ∼4.5 unexpectedly resulted in a nearly two-fold reduction in cellular lipid accumulation compared to the parental control strain, a sign of cellular metabolic drain. Such metabolic drain was ameliorated when grown in media with a high C/N ratio of 59 having a higher glucose utilization rate that led to approximately twofold more cell mass and threefold more lipid production per liter culture compared to parental control strain, suggesting cross-talk between cellulase and lipid production, both of which involve the endoplasmic reticulum (ER). Most importantly, we found that the chemical chaperone, trimethylamine N-oxide dihydride increased glucose utilization, cell mass and total lipid titer in the transformants, suggesting further amelioration of the metabolic drain. This is the first study examining lipid production in cellulase-expressing Y. lipolytica strains under various C/N ratio media and with a chemical chaperone highlighting the metabolic complexity for developing robust, cellulolytic and lipogenic yeast strains.

Engineering a Glucosamine-6-phosphate Responsive glmS Ribozyme Switch Enables Dynamic Control of Metabolic Flux in Bacillus subtilis for Overproduction of N-Acetylglucosamine

Bacillus subtilis is a typical industrial microorganism and is widely used in industrial biotechnology, particularly for nutraceutical production. There are many studies on the static metabolic engineering of B. subtilis, whereas there are few reports on dynamic metabolic engineering due to the lack of appropriate elements. Here, we established a dynamic reprogramming strategy for reconstructing metabolic networks in B. subtilis, using a typical nutraceutical, N-acetylglucosamine (GlcNAc), as a model product and the glmS (encoding glucosamine-6-phosphate synthase) ribozyme as an engineering element. First, a trp terminator was introduced to effectively release the glmS ribozyme feedback inhibition. Further, we engineered the native glucosamine-6-phosphate (GlcN6P) responsive glmS ribozyme switch to dynamically control the metabolic flux in B. subtilis for overproduction of GlcNAc. With GlcN6P as a ligand, the native sensor glmS ribozyme is integrated at the 5'- of phosphoglucosamine mutase and 6-phosphofructokinase genes to decrease the flux dynamically toward the peptidoglycan synthesis and glycolysis pathway, respectively. The glmS ribozyme mutant M5 ( glmS ribozyme cleavage site AG → GG) with decreased ribozyme activity is integrated at the 5'- of glucose-6-phosphate isomerase gene to increase the flux dynamically toward the GlcNAc synthesis pathway. This strategy increased the GlcNAc titer from 9.24 to 18.45 g/L, and the specific GlcNAc productivity from 0.53 to 1.21 g GlcNAc/g cell. Since GlcN6P is involved in the biosynthesis of various products, here the developed strategy for multiple target dynamic engineering of metabolic pathways can be generally used in B. subtilis and other industrial microbes for chemical production.

Production of optically pure L(+)-lactic acid from waste plywood chips using an isolated thermotolerant Enterococcus faecalis SI at a pilot scale

Utilization of renewable and low-cost lignocellulosic wastes has received major focus in industrial lactic acid production. The use of high solid loadings in biomass pretreatment potentially offers advantages over low solid loadings including higher lactic acid concentration with decreased production and capital costs. In this study, an isolated Enterococcus faecalis SI with optimal temperature 42 °C was used to produce optically pure L-lactic acid (> 99%) from enzyme-saccharified hydrolysates of acid-impregnated steam explosion (AISE)-treated plywood chips. The L-lactic acid production increased by 10% at 5 L scale compared to the similar fermentation scheme reported by Wee et al. The fermentation with a high solid loading of 20% and 35% (w/v) AISE-pretreated plywood chips had been successfully scaled up to process development unit scale (100 L) and pilot scale (9 m³), respectively. This is the first report of pilot-scale lignocellulosic lactic acid fermentation by E. faecalis with high lactic acid titer (nearly 92 g L^-1) and yield (0.97 kg kg^-1). Therefore, large-scale L-lactic acid production by E. faecalis SI shows the potential application for industries.

Modular Ligation Extension of Guide RNA Operons (LEGO) for Multiplexed dCas9 Regulation of Metabolic Pathways in Saccharomyces cerevisiae

Metabolic engineering typically utilizes a suboptimal step-wise gene target optimization approach to parse a highly connected and regulated cellular metabolism. While the endonuclease-null CRISPR/Cas system has enabled gene expression perturbations without genetic modification, it has been mostly limited to small sets of gene targets in eukaryotes due to inefficient methods to assemble and express large sgRNA operons. In this work, we develop a TEF1p-tRNA expression system and demonstrate that the use of tRNAs as splicing elements flanking sgRNAs provides higher efficiency than both Pol III and ribozyme-based expression across a variety of single sgRNA and multiplexed contexts. Next, we devise and validate a scheme to allow modular construction of tRNA-sgRNA (TST) operons using an iterative Type IIs digestion/ligation extension approach, termed CRISPR-Ligation Extension of sgRNA Operons (LEGO). This approach enables facile construction of large TST operons. We demonstrate this utility by constructing a metabolic rewiring prototype for 2,3-butanediol production in 2 distinct yeast strain backgrounds. These results demonstrate that our approach can act as a surrogate for traditional genetic modification on a much shorter design-cycle timescale.

Engineering Yarrowia lipolytica for the production of cyclopropanated fatty acids

Traditional synthesis of biodiesel competes with food sources and has limitations with storage, particularly due to limited oxidative stability. Microbial synthesis of lipids provides a platform to produce renewable fuel with improved properties from various renewable carbon sources. Specifically, biodiesel properties can be improved through the introduction of a cyclopropane ring in place of a double bond. In this study, we demonstrate the production of C19 cyclopropanated fatty acids in the oleaginous yeast Yarrowia lipolytica through the heterologous expression of the Escherichia coli cyclopropane fatty acid synthase. Ultimately, we establish a strain capable of 3.03 ± 0.26 g/L C19 cyclopropanated fatty acid production in bioreactor fermentation where this functionalized lipid comprises over 32% of the total lipid pool. This study provides a demonstration of the flexibility of lipid metabolism in Y. lipolytica to produce specialized fatty acids.

Metabolic engineering in the host Yarrowia lipolytica

The nonconventional, oleaginous yeast, Yarrowia lipolytica is rapidly emerging as a valuable host for the production of a variety of both lipid and nonlipid chemical products. While the unique genetics of this organism pose some challenges, many new metabolic engineering tools have emerged to facilitate improved genetic manipulation in this host. This review establishes a case for Y. lipolytica as a premier metabolic engineering host based on innate metabolic capacity, emerging synthetic tools, and engineering examples. The metabolism underlying the lipid accumulation phenotype of this yeast as well as high flux through acyl-CoA precursors and the TCA cycle provide a favorable metabolic environment for expression of relevant heterologous pathways. These properties allow Y. lipolytica to be successfully engineered for the production of both native and nonnative lipid, organic acid, sugar and acetyl-CoA derived products. Finally, this host has unique metabolic pathways enabling growth on a wide range of carbon sources, including waste products. The expansion of carbon sources, together with the improvement of tools as highlighted here, have allowed this nonconventional organism to act as a cellular factory for valuable chemicals and fuels.

Synthetic Biology Expands the Industrial Potential of Yarrowia lipolytica

The oleaginous yeast Yarrowia lipolytica is quickly emerging as the most popular non-conventional (i.e., non-model organism) yeast in the bioproduction field. With a high propensity for flux through tricarboxylic acid (TCA) cycle intermediates and biological precursors such as acetyl-CoA and malonyl-CoA, this host is especially well suited to meet our industrial chemical production needs. Recent progress in synthetic biology tool development has greatly enhanced our ability to rewire this organism, with advances in genetic component design, CRISPR technologies, and modular cloning strategies. In this review we investigate recent developments in metabolic engineering and describe how the new tools being developed help to realize the full industrial potential of this host. Finally, we conclude with our vision of the developments that will be necessary to enhance future engineering efforts.

Largely enhanced bioethanol production through the combined use of lignin-modified sugarcane and xylose fermenting yeast strain

The recalcitrant structure of lignocellulosic biomass is a major barrier in efficient biomass-to-ethanol bioconversion processes. The combination of feedstock engineering via modification in the lignin synthesis pathway of sugarcane and co-fermentation of xylose and glucose with a recombinant xylose utilizing yeast strain produced 148% more ethanol compared to that of the wild type biomass and control strain. The lignin reduced biomass led to a substantially increased release of fermentable sugars (glucose and xylose). The engineered yeast strain efficiently co-utilized glucose and xylose for fermentation, elevating ethanol yields. In this study, it was experimentally demonstrated that the combined efforts of engineering both feedstock and microorganisms largely enhances the bioconversion of lignocellulosic feedstock to bioethanol. This strategy will significantly improve the economic feasibility of lignocellulosic biofuels production.

Developing a piggyBac Transposon System and Compatible Selection Markers for Insertional Mutagenesis and Genome Engineering in Yarrowia lipolytica

Yarrowia lipolytica is a non-conventional yeast of interest to the biotechnology industry. However, the physiology, metabolism, and genetic regulation of Y. lipolytica diverge significantly from more well-studied and characterized yeasts such as Saccharomyces cerevisiae. To develop additional genetic tools for this industrially relevant host, the piggyBac transposon system to enable efficient generation of genome-wide insertional mutagenesis libraries and introduction of scarless, footprint-free genomic modifications in Y. lipolytica. Specifically, we demonstrate piggyBac transposition in Y. lipolytica, and then use the approach to screen transposon insertion libraries for rapid isolation of mutations that confer altered canavanine resistance, pigment formation, and neutral lipid accumulation. We also develop a variety of piggyBac compatible selection markers for footprint-free genome engineering, including a novel dominant marker cassette (Escherichia coli guaB) for effective Y. lipolytica selection using mycophenolic acid. We utilize these marker cassettes to construct a piggyBac vector set that allows for auxotrophic selection (uracil or tryptophan biosynthesis) or dominant selection (hygromycin, nourseothricin, chlorimuron ethyl, or mycophenolic acid resistance) and subsequent marker excision. These new genetic tools and techniques will help to facilitate and accelerate the engineering of Y. lipolytica strains for efficient and sustainable production of a wide variety of small molecules and proteins.

T7 Polymerase Expression of Guide RNAs in vivo Allows Exportable CRISPR-Cas9 Editing in Multiple Yeast Hosts

Efficient guide RNA expression often limits CRISPR-Cas9 implementation in new hosts. To address this limitation in fungal systems, we demonstrate the utility of a T7 polymerase system to effectively express sgRNAs. Initially, we developed a methodology in Saccharomyces cerevisiae using a modified version of the T7 P266L mutant polymerase with an SV40 nuclear localization signal to allow guide RNA expression immediately downstream of a T7 promoter. To improve targeting efficiency, guide RNA design was found to be tolerant to three mismatches or up to three additional bases appended to the 5' end. The addition of three guanines to a T7-based guide RNA improved guide RNA expression 80-fold and achieved transcriptional output similar to the strong Pol III snr52 promoter. Resulting gene editing and dCas9-guided gene regulation with a T7-based guide RNA was on par with the commonly used snr52 system in S. cerevisiae. Finally, 96% and 60% genome editing efficiencies were achieved in Kluyveromyces lactis and Yarrowia lipolytica respectively with minimal optimization of this system. Thus, T7-based expression of sgRNAs offers an orthogonal method for implementing CRISPR systems in fungal systems.

Condition-specific promoter activities in Saccharomyces cerevisiae

BACKGROUND

Saccharomyces cerevisiae is widely studied for production of biofuels and biochemicals. To improve production efficiency under industrially relevant conditions, coordinated expression of multiple genes by manipulating promoter strengths is an efficient approach. It is known that gene expression is highly dependent on the practically used environmental conditions and is subject to dynamic changes. Therefore, investigating promoter activities of S. cerevisiae under different culture conditions in different time points, especially under stressful conditions is of great importance.

RESULTS

In this study, the activities of various promoters in S. cerevisiae under stressful conditions and in the presence of xylose were characterized using yeast enhanced green fluorescent protein (yEGFP) as a reporter. The stresses include toxic levels of acetic acid and furfural, and high temperature, which are related to fermentation of lignocellulosic hydrolysates. In addition to investigating eight native promoters, the synthetic hybrid promoter P_3xC-TEF1 was also evaluated. The results revealed that P _TDH3 and the synthetic promoter P_3xC-TEF1 showed the highest strengths under almost all the conditions. Importantly, these two promoters also exhibited high stabilities throughout the cultivation. However, the strengths of P _ADH1 and P _PGK1 , which are generally regarded as 'constitutive' promoters, decreased significantly under certain conditions, suggesting that cautions should be taken to use such constitutive promoters to drive gene expression under stressful conditions. Interestingly, P _HSP12 and P _HSP26 were able to response to both high temperature and acetic acid stress. Moreover, P _HSP12 also led to moderate yEGFP expression when xylose was used as the sole carbon source, indicating that this promoter could be used for inducing proper gene expression for xylose utilization.

CONCLUSION

The results here revealed dynamic changes of promoter activities in S. cerevisiae throughout batch fermentation in the presence of inhibitors as well as using xylose. These results provide insights in selection of promoters to construct S. cerevisiae strains for efficient bioproduction under practical conditions. Our results also encouraged applications of synthetic promoters with high stability for yeast strain development.

Expanding beyond canonical metabolism: Interfacing alternative elements, synthetic biology, and metabolic engineering

Metabolic engineering offers an exquisite capacity to produce new molecules in a renewable manner. However, most industrial applications have focused on only a small subset of elements from the periodic table, centered around carbon biochemistry. This review aims to illustrate the expanse of chemical elements that can currently (and potentially) be integrated into useful products using cellular systems. Specifically, we describe recent advances in expanding the cellular scope to include the halogens, selenium and the metalloids, and a variety of metal incorporations. These examples range from small molecules, heteroatom-linked uncommon elements, and natural products to biomining and nanotechnology applications. Collectively, this review covers the promise of an expanded range of elemental incorporations and the future impacts it may have on biotechnology.

Yeast Terminator Function Can Be Modulated and Designed on the Basis of Predictions of Nucleosome Occupancy

The design of improved synthetic parts is a major goal of synthetic biology. Mechanistically, nucleosome occupancy in the 3' terminator region of a gene has been found to correlate with transcriptional expression. Here, we seek to establish a predictive relationship between terminator function and predicted nucleosome positioning to design synthetic terminators in the yeast Saccharomyces cerevisiae. In doing so, terminators improved net protein output from these expression cassettes nearly 4-fold over their original sequence with observed increases in termination efficiency to 96%. The resulting terminators were indeed depleted of nucleosomes on the basis of mapping experiments. This approach was successfully applied to synthetic, de novo, and native terminators. The mode of action of these modifications was mainly through increased termination efficiency, rather than half-life increases, perhaps suggesting a role in improved mRNA maturation. Collectively, these results suggest that predicted nucleosome depletion can be used as a heuristic approach for improving terminator function, though the underlying mechanism remains to be shown.

Enabling Graded and Large-Scale Multiplex of Desired Genes Using a Dual-Mode dCas9 Activator in Saccharomyces cerevisiae

Standard approaches for dCas9-based modification of gene expression are limited in the ability to multiplex targets, establish streamlined cassettes, and utilize commonly studied Pol II promoters. In this work, we repurpose the dCas9-VPR activator to act as a dual-mode activator/repressor that can be programmed solely on the basis of target position at gene loci. Furthermore, we implement this approach using a streamlined Pol II-ribozyme system that allows expression of many sgRNAs from a single transcript. By "stepping" dCas9-VPR within the promoter region and ORF we create graded activation and repression (respectively) of target genes, allowing precise control over multiplexed gene modulation. Expression from the Pol II system increased the net amount of sgRNA production in cells by 3.88-fold relative to the Pol III SNR52 promoter, leading to a significant improvement in dCas9-VPR repression strength. Finally, we utilize our Pol II system to create galactose-inducible switching of gene expression states and multiplex constructs capable of modulating up to 4 native genes from a single vector. Our approach represents a significant step toward minimizing DNA required to assemble CRISPR systems in eukaryotes while enhancing the efficacy (greater repression strength), scale (more sgRNAs), and scope (inducibility) of dCas9-mediated gene regulation.