Trade-Offs in Biosensor Optimization for Dynamic Pathway Engineering

A hybrid in silico/in-cell controller that handles process-model mismatches using intracellular biosensing

The discrepancy between model predictions and actual processes, known as process-model mismatch (PMM), remains a substantial challenge in bioprocess optimization. We previously introduced a hybrid in silico/in-cell controller (HISICC) that combines model-based optimization with cell-based feedback to address this problem. Here, we extended this approach to regulate a key enzyme level using intracellular biosensing. The extended HISICC was implemented using an Escherichia coli strain engineered for fatty acid production (FA3). This strain contains a genetically encoded feedback controller that decelerates the expression of acetyl-CoA carboxylase (ACC) in response to malonyl-CoA synthesized through the enzymatic reaction. We modeled FA3 to allow the HISICC to optimize an inducer input that accelerates the enzyme expression. Simulations showed that the HISICC slowed the unexpectedly rapid accumulation of ACC resulting from PMMs before it reached cytotoxic levels, thereby improving fatty acid yields. These results highlight the potential of our approach, particularly in cases where monitoring intracellular biomolecules is required to handle PMMs.

Optimizing HMG-CoA Synthase Expression for Enhanced Limonene Production in Escherichia coli through Temporal Transcription Modulation Using Optogenetics

Overexpression of a single enzyme in a multigene heterologous pathway may be out of balance with the other enzymes in the pathway, leading to accumulated toxic intermediates, imbalanced carbon flux, reduced productivity of the pathway, or an inhibited growth phenotype. Therefore, optimal, balanced, and synchronized expression levels of enzymes in a particular metabolic pathway is critical to maximize production of desired compounds while maintaining cell fitness in a growing culture. Furthermore, the optimal intracellular concentration of an enzyme is determined by the expression strength, specific timing/duration, and degradation rate of the enzyme. Here, we modulated the intracellular concentration of a key enzyme, namely HMG-CoA synthase (HMGS), in the heterologous mevalonate pathway by tuning its expression level and period of transcription to enhance limonene production in Escherichia coli. Facilitated by the tuned blue-light inducible BLADE/pBad system, we observed that limonene production was highest (160 mg/L) with an intermediate transcription level of HMGS from moderate light illumination (41 au, 150 s ON/150 s OFF) throughout the growth. Owing to the easy penetration and removal of blue-light illumination from the growing culture which is hard to obtain using conventional chemical-based induction, we further explored different induction patterns of HMGS under strong light illumination (2047 au, 300 s ON) for different durations along the growth phases. We identified a specific timing of HMGS expression in the log phase (3-9 h) that led to optimal limonene production (200 mg/L). This is further supported by a mathematical model that predicts several periods of blue-light illumination (3-9 h, 0-9 h, 3-12 h, 0-12 h) to achieve an optimal expression level of HMGS that maximizes limonene production and maintains cell fitness. Compared to moderate and prolonged transcription (41 au, 150 s ON/150 s OFF, 0-73 h), strong but time-limited transcription (2047 au, 300 s ON, 3-9 h) of HMGS could maintain its optimal intracellular concentration and further increased limonene production up to 92% (250 mg/L) in the longer incubation (up to 73 h) without impacting cell fitness. This work has provided new insight into the "right amount" and "just-in-time" expression of a critical metabolite enzyme in the upper module of the mevalonate pathway using optogenetics. This study would complement previous findings in modulating HMGS expression and potentially be applicable to heterologous production of other terpenoids in E. coli.

Delaying production with prokaryotic inducible expression systems

BACKGROUND

Engineering bacteria with the purpose of optimizing the production of interesting molecules often leads to a decrease in growth due to metabolic burden or toxicity. By delaying the production in time, these negative effects on the growth can be avoided in a process called a two-stage fermentation.

MAIN TEXT

During this two-stage fermentation process, the production stage is only activated once sufficient cell mass is obtained. Besides the possibility of using external triggers, such as chemical molecules or changing fermentation parameters to induce the production stage, there is a renewed interest towards autoinducible systems. These systems, such as quorum sensing, do not require the extra interference with the fermentation broth to start the induction. In this review, we discuss the different possibilities of both external and autoinduction methods to obtain a two-stage fermentation. Additionally, an overview is given of the tuning methods that can be applied to optimize the induction process. Finally, future challenges and prospects of (auto)inducible expression systems are discussed.

CONCLUSION

There are numerous methods to obtain a two-stage fermentation process each with their own advantages and disadvantages. Even though chemically inducible expression systems are well-established, an increasing interest is going towards autoinducible expression systems, such as quorum sensing. Although these newer techniques cannot rely on the decades of characterization and applications as is the case for chemically inducible promoters, their advantages might lead to a shift in future inducible expression systems.

Cell factories for biosynthesis of D-glucaric acid: a fusion of static and dynamic strategies

D-glucaric acid is an important organic acid with numerous applications in therapy, food, and materials, contributing significantly to its substantial market value. The biosynthesis of D-glucaric acid (GA) from renewable sources such as glucose has garnered significant attention due to its potential for sustainable and cost-effective production. This review summarizes the current understanding of the cell factories for GA production in different chassis strains, from static to dynamic control strategies for regulating their metabolic networks. We highlight recent advances in the optimization of D-glucaric acid biosynthesis, including metabolic dynamic control, alternative feedstocks, metabolic compartments, and so on. Additionally, we compare the differences between different chassis strains and discuss the challenges that each chassis strain must overcome to achieve highly efficient GA productions. In this review, the processes of engineering a desirable cell factory for highly efficient GA production are just like an epitome of metabolic engineering of strains for chemical biosynthesis, inferring general trends for industrial chassis strain developments.

Cross-regulation between proteome reallocation and metabolic flux redistribution governs bacterial growth transition kinetics

Bacteria need to adjust their metabolism and protein synthesis simultaneously to adapt to changing nutrient conditions. It's still a grand challenge to predict how cells coordinate such adaptation due to the cross-regulation between the metabolic fluxes and the protein synthesis. Here we developed a dynamic Constrained Allocation Flux Balance Analysis method (dCAFBA), which integrates flux-controlled proteome allocation and protein limited flux balance analysis. This framework can predict the redistribution dynamics of metabolic fluxes without requiring detailed enzyme parameters. We reveal that during nutrient up-shifts, the calculated metabolic fluxes change in agreement with experimental measurements of enzyme protein dynamics. During nutrient down-shifts, we uncover a switch of metabolic bottleneck from carbon uptake proteins to metabolic enzymes, which disrupts the coordination between metabolic flux and their enzyme abundance. Our method provides a quantitative framework to investigate cellular metabolism under varying environments and reveals insights into bacterial adaptation strategies.

Production of D-glucaric acid with phosphoglucose isomerase-deficient Saccharomyces cerevisiae

D-Glucaric acid is a potential biobased platform chemical. Previously mainly Escherichia coli, but also the yeast Saccharomyces cerevisiae, and Pichia pastoris, have been engineered for conversion of D-glucose to D-glucaric acid via myo-inositol. One reason for low yields from the yeast strains is the strong flux towards glycolysis. Thus, to decrease the flux of D-glucose to biomass, and to increase D-glucaric acid yield, the four step D-glucaric acid pathway was introduced into a phosphoglucose isomerase deficient (Pgi1p-deficient) Saccharomyces cerevisiae strain. High D-glucose concentrations are toxic to the Pgi1p-deficient strains, so various feeding strategies and use of polymeric substrates were studied. Uniformly labelled 13C-glucose confirmed conversion of D-glucose to D-glucaric acid. In batch bioreactor cultures with pulsed D-fructose and ethanol provision 1.3 g D-glucaric acid L-1 was produced. The D-glucaric acid titer (0.71 g D-glucaric acid L-1) was lower in nitrogen limited conditions, but the yield, 0.23 g D-glucaric acid [g D-glucose consumed]-1, was among the highest that has so far been reported from yeast. Accumulation of myo-inositol indicated that myo-inositol oxygenase activity was limiting, and that there would be potential to even higher yield. The Pgi1p-deficiency in S. cerevisiae provides an approach that in combination with other reported modifications and bioprocess strategies would promote the development of high yield D-glucaric acid yeast strains.

Applications of artificial intelligence and machine learning in dynamic pathway engineering

Dynamic pathway engineering aims to build metabolic production systems embedded with intracellular control mechanisms for improved performance. These control systems enable host cells to self-regulate the temporal activity of a production pathway in response to perturbations, using a combination of biosensors and feedback circuits for controlling expression of heterologous enzymes. Pathway design, however, requires assembling together multiple biological parts into suitable circuit architectures, as well as careful calibration of the function of each component. This results in a large design space that is costly to navigate through experimentation alone. Methods from artificial intelligence (AI) and machine learning are gaining increasing attention as tools to accelerate the design cycle, owing to their ability to identify hidden patterns in data and rapidly screen through large collections of designs. In this review, we discuss recent developments in the application of machine learning methods to the design of dynamic pathways and their components. We cover recent successes and offer perspectives for future developments in the field. The integration of AI into metabolic engineering pipelines offers great opportunities to streamline design and discover control systems for improved production of high-value chemicals.

In Vivo Detection of Low Molecular Weight Platform Chemicals and Environmental Contaminants by Genetically Encoded Biosensors

Genetically encoded biosensor systems operating in living cells are versatile, cheap, and transferable tools for the detection and quantification of a broad range of small molecules. This review presents state-of-the-art biosensor designs and assemblies, featuring transcription factor-, riboswitch-, and enzyme-coupled devices, highly engineered fluorescent probes, and emerging two-component systems. Importantly, (bioinformatic-assisted) strategies to resolve contextual issues, which cause biosensors to miss performance criteria in vivo, are highlighted. The optimized biosensing circuits can be used to monitor chemicals of low molecular mass (<200 g mol-1) and physicochemical properties that challenge conventional chromatographical methods with high sensitivity. Examples herein include but are not limited to formaldehyde, formate, and pyruvate as immediate products from (synthetic) pathways for the fixation of carbon dioxide (CO2), industrially important derivatives like small- and medium-chain fatty acids and biofuels, as well as environmental toxins such as heavy metals or reactive oxygen and nitrogen species. Lastly, this review showcases biosensors capable of assessing the biosynthesis of platform chemicals from renewable resources, the enzymatic degradation of plastic waste, or the bioadsorption of highly toxic chemicals from the environment. These applications offer new biosensor-based manufacturing, recycling, and remediation strategies to tackle current and future environmental and socioeconomic challenges including the wastage of fossil fuels, the emission of greenhouse gases like CO2, and the pollution imposed on ecosystems and human health.

Bayesian Optimization for Design of Multiscale Biological Circuits

Recent advances in synthetic biology have enabled the construction of molecular circuits that operate across multiple scales of cellular organization, such as gene regulation, signaling pathways, and cellular metabolism. Computational optimization can effectively aid the design process, but current methods are generally unsuited for systems with multiple temporal or concentration scales, as these are slow to simulate due to their numerical stiffness. Here, we present a machine learning method for the efficient optimization of biological circuits across scales. The method relies on Bayesian optimization, a technique commonly used to fine-tune deep neural networks, to learn the shape of a performance landscape and iteratively navigate the design space toward an optimal circuit. This strategy allows the joint optimization of both circuit architecture and parameters, and provides a feasible approach to solve a highly nonconvex optimization problem in a mixed-integer input space. We illustrate the applicability of the method on several gene circuits for controlling biosynthetic pathways with strong nonlinearities, multiple interacting scales, and using various performance objectives. The method efficiently handles large multiscale problems and enables parametric sweeps to assess circuit robustness to perturbations, serving as an efficient in silico screening method prior to experimental implementation.

Applications and Tuning Strategies for Transcription Factor-Based Metabolite Biosensors

Transcription factor (TF)-based biosensors are widely used for the detection of metabolites and the regulation of cellular pathways in response to metabolites. Several challenges hinder the direct application of TF-based sensors to new hosts or metabolic pathways, which often requires extensive tuning to achieve the optimal performance. These tuning strategies can involve transcriptional or translational control depending on the parameter of interest. In this review, we highlight recent strategies for engineering TF-based biosensors to obtain the desired performance and discuss additional design considerations that may influence a biosensor's performance. We also examine applications of these sensors and suggest important areas for further work to continue the advancement of small-molecule biosensors.

Precision engineering of biological function with large-scale measurements and machine learning

As synthetic biology expands and accelerates into real-world applications, methods for quantitatively and precisely engineering biological function become increasingly relevant. This is particularly true for applications that require programmed sensing to dynamically regulate gene expression in response to stimuli. However, few methods have been described that can engineer biological sensing with any level of quantitative precision. Here, we present two complementary methods for precision engineering of genetic sensors: in silico selection and machine-learning-enabled forward engineering. Both methods use a large-scale genotype-phenotype dataset to identify DNA sequences that encode sensors with quantitatively specified dose response. First, we show that in silico selection can be used to engineer sensors with a wide range of dose-response curves. To demonstrate in silico selection for precise, multi-objective engineering, we simultaneously tune a genetic sensor's sensitivity (EC50) and saturating output to meet quantitative specifications. In addition, we engineer sensors with inverted dose-response and specified EC50. Second, we demonstrate a machine-learning-enabled approach to predictively engineer genetic sensors with mutation combinations that are not present in the large-scale dataset. We show that the interpretable machine learning results can be combined with a biophysical model to engineer sensors with improved inverted dose-response curves.

Accuracy and data efficiency in deep learning models of protein expression

Synthetic biology often involves engineering microbial strains to express high-value proteins. Thanks to progress in rapid DNA synthesis and sequencing, deep learning has emerged as a promising approach to build sequence-to-expression models for strain optimization. But such models need large and costly training data that create steep entry barriers for many laboratories. Here we study the relation between accuracy and data efficiency in an atlas of machine learning models trained on datasets of varied size and sequence diversity. We show that deep learning can achieve good prediction accuracy with much smaller datasets than previously thought. We demonstrate that controlled sequence diversity leads to substantial gains in data efficiency and employed Explainable AI to show that convolutional neural networks can finely discriminate between input DNA sequences. Our results provide guidelines for designing genotype-phenotype screens that balance cost and quality of training data, thus helping promote the wider adoption of deep learning in the biotechnology sector.

Advances in engineering and optimization of transcription factor-based biosensors for plug-and-play small molecule detection

Transcription factor (TF)-based biosensors have been applied in biotechnology for a variety of functions, including protein engineering, dynamic control, environmental detection, and point-of-care diagnostics. Such biosensors are promising analytical tools due to their wide range of detectable ligands and modular nature. However, designing biosensors tailored for applications of interest with the desired performance parameters, including ligand specificity, remains challenging. Biosensors often require significant engineering and tuning to meet desired specificity, sensitivity, dynamic range, and operating range parameters. Another limitation is the orthogonality of biosensors across hosts, given the role of the cellular context. Here, we describe recent advances and examples in the engineering and optimization of TF-based biosensors for plug-and-play small molecule detection. We highlight novel developments in TF discovery and biosensor design, TF specificity engineering, and biosensor tuning, with emphasis on emerging computational methods.

The Growth Dependent Design Constraints of Transcription-Factor-Based Metabolite Biosensors

Metabolite biosensors based on metabolite-responsive transcription factors are key synthetic biology components for sensing and precisely controlling cellular metabolism. Biosensors are often designed under laboratory conditions but are deployed in applications where cellular growth rate differs drastically from its initial characterization. Here we asked how growth rate impacts the minimum and maximum biosensor outputs and the dynamic range, which are key metrics of biosensor performance. Using LacI, TetR, and FadR-based biosensors in Escherichia coli as models, we find that the dynamic range of different biosensors have different growth rate dependencies. We developed a kinetic model to explore how tuning biosensor parameters impact the dynamic range growth rate dependence. Our modeling and experimental results revealed that the effects to dynamic range and its growth rate dependence are often coupled, and the metabolite transport mechanisms shape the dynamic range-growth rate response. This work provides a systematic understanding of biosensor performance under different growth rates, which will be useful for predicting biosensor behavior in broad synthetic biology and metabolic engineering applications.