High-throughput screening, next generation sequencing and machine learning: advanced methods in enzyme engineering

Functionalized Polysaccharides Improve Sensitivity of Tyramide/Peroxidase Proximity Labeling Assays through Electrostatic Interactions

High-throughput assays that efficiently link genotype and phenotype with high fidelity are key to successful enzyme engineering campaigns. Among these assays, the tyramide/peroxidase proximity labeling method converts the product of an enzymatic reaction of a surface expressed enzyme to a highly reactive fluorescent radical, which labels the cell surface. In this context, maintaining the proximity of the readout reagents to the cell surface is crucial to prevent crosstalk and ensure that short-lived radical species react before diffusing away. Here, we investigated improvements in tyramide/peroxidase proximity labeling for enzyme screening. We modified chitosan (Cs) chains with horseradish peroxidase (HRP) and evaluated the effects of these conjugates on the efficiency of proximity labeling reactions on yeast cells displaying d-amino acid oxidase. By tethering HRP to chitosan through different chemical approaches, we localized the auxiliary enzyme close to the cell surface and enhanced the sensitivity of tyramide-peroxidase labeling reactions. We found that immobilizing HRP onto chitosan through a 5 kDa PEG linker improved labeling sensitivity by over 3.5-fold for substrates processed with a low turnover rate (e.g., d-lysine), while the sensitivity of the labeling for high activity substrates (e.g., d-alanine) was enhanced by over 0.6-fold. Such improvements in labeling efficiency broaden the range of enzymes and conditions that can be studied and screened by tyramide/peroxidase proximity labeling.

Evolutionary Probability and Stacked Regressions Enable Data-Driven Protein Engineering with Minimized Experimental Effort

Protein engineering through directed evolution and (semi)rational approaches is routinely applied to optimize protein properties for a broad range of applications in industry and academia. The multitude of possible variants, combined with limited screening throughput, hampers efficient protein engineering. Data-driven strategies have emerged as a powerful tool to model the protein fitness landscape that can be explored in silico, significantly accelerating protein engineering campaigns. However, such methods require a certain amount of data, which often cannot be provided, to generate a reliable model of the fitness landscape. Here, we introduce MERGE, a method that combines direct coupling analysis (DCA) and machine learning (ML). MERGE enables data-driven protein engineering when only limited data are available for training, typically ranging from 50 to 500 labeled sequences. Our method demonstrates remarkable performance in predicting a protein's fitness value and rank based on its sequence across diverse proteins and properties. Notably, MERGE outperforms state-of-the-art methods when only small data sets are available for modeling, requiring fewer computational resources, and proving particularly promising for protein engineers who have access to limited amounts of data.

Navigating the landscape of enzyme design: from molecular simulations to machine learning

Global environmental issues and sustainable development call for new technologies for fine chemical synthesis and waste valorization. Biocatalysis has attracted great attention as the alternative to the traditional organic synthesis. However, it is challenging to navigate the vast sequence space to identify those proteins with admirable biocatalytic functions. The recent development of deep-learning based structure prediction methods such as AlphaFold2 reinforced by different computational simulations or multiscale calculations has largely expanded the 3D structure databases and enabled structure-based design. While structure-based approaches shed light on site-specific enzyme engineering, they are not suitable for large-scale screening of potential biocatalysts. Effective utilization of big data using machine learning techniques opens up a new era for accelerated predictions. Here, we review the approaches and applications of structure-based and machine-learning guided enzyme design. We also provide our view on the challenges and perspectives on effectively employing enzyme design approaches integrating traditional molecular simulations and machine learning, and the importance of database construction and algorithm development in attaining predictive ML models to explore the sequence fitness landscape for the design of admirable biocatalysts.

Droplet Microfluidics for High-Throughput Screening and Directed Evolution of Biomolecules

Directed evolution is a powerful technique for creating biomolecules such as proteins and nucleic acids with tailor-made properties for therapeutic and industrial applications by mimicking the natural evolution processes in the laboratory. Droplet microfluidics improved classical directed evolution by enabling time-consuming and laborious steps in this iterative process to be performed within monodispersed droplets in a highly controlled and automated manner. Droplet microfluidic chips can generate, manipulate, and sort individual droplets at kilohertz rates in a user-defined microchannel geometry, allowing new strategies for high-throughput screening and evolution of biomolecules. In this review, we discuss directed evolution studies in which droplet-based microfluidic systems were used to screen and improve the functional properties of biomolecules. We provide a systematic overview of basic on-chip fluidic operations, including reagent mixing by merging continuous fluid streams and droplet pairs, reagent addition by picoinjection, droplet generation, droplet incubation in delay lines, chambers and hydrodynamic traps, and droplet sorting techniques. Various microfluidic strategies for directed evolution using single and multiple emulsions and biomimetic materials (giant lipid vesicles, microgels, and microcapsules) are highlighted. Completely cell-free microfluidic-assisted in vitro compartmentalization methods that eliminate the need to clone DNA into cells after each round of mutagenesis are also presented.

Droplet-based cell-laden microgels for high-throughput analysis

Cell-laden droplet microfluidics has revolutionized bulk biochemical analysis by offering compartmentalized microreactors for individual cells, but downstream operations of regular aqueous droplets are limited. Hydrogel matrix can provide a rigid scaffold for long-term culture of eukaryotic and prokaryotic cells, and can support several manipulations, facilitating subsequent high-throughput analysis of cellular heterogeneity.

Advanced strategies in high-throughput droplet screening for enzyme engineering

Enzymes, as biocatalysts, play a cumulatively important role in environmental purification and industrial production of chemicals and pharmaceuticals. However, natural enzymes are limited by their physiological properties in practice, which need to be modified driven by requirements. Screening and isolating certain enzyme variants or ideal industrial strains with high yielding of target product enzymes is one of the main directions of enzyme engineering research. Droplet-based high-throughput screening (DHTS) technology employs massive monodisperse emulsion droplets as microreactors to achieve single strain encapsulation, as well as continuous monitoring for the inside mutant library. It can effectively sort out strains or enzymes with desired characteristics, offering a throughput of 108 events per hour. Much of the early literature focused on screening various engineered strains or designing signalling sorting strategies based on DHTS technology. However, the field of enzyme engineering lacks a comprehensive overview of advanced methods for microfluidic droplets and their cutting-edge developments in generation and manipulation. This review emphasizes the advanced strategies and frontiers of microfluidic droplet generation and manipulation facilitating enzyme engineering development. We also introduce design for various screening signals that cooperate with DHTS and devote to enzyme engineering.

Understanding activity-stability tradeoffs in biocatalysts by enzyme proximity sequencing

Understanding the complex relationships between enzyme sequence, folding stability and catalytic activity is crucial for applications in industry and biomedicine. However, current enzyme assay technologies are limited by an inability to simultaneously resolve both stability and activity phenotypes and to couple these to gene sequences at large scale. Here we present the development of enzyme proximity sequencing, a deep mutational scanning method that leverages peroxidase-mediated radical labeling with single cell fidelity to dissect the effects of thousands of mutations on stability and catalytic activity of oxidoreductase enzymes in a single experiment. We use enzyme proximity sequencing to analyze how 6399 missense mutations influence folding stability and catalytic activity in a D-amino acid oxidase from Rhodotorula gracilis. The resulting datasets demonstrate activity-based constraints that limit folding stability during natural evolution, and identify hotspots distant from the active site as candidates for mutations that improve catalytic activity without sacrificing stability. Enzyme proximity sequencing can be extended to other enzyme classes and provides valuable insights into biophysical principles governing enzyme structure and function.

High-Throughput Screening Techniques for the Selection of Thermostable Enzymes

The acquisition of a thermostable enzyme is an indispensable prerequisite for its successful implementation in industrial applications and the development of novel functionalities. Various protein engineering approaches, including rational design, semirational design, and directed evolution, have been employed to enhance thermostability. However, all of these approaches require sensitive and reliable high-throughput screening (HTS) technologies to efficiently and rapidly identify variants with improved properties. While numerous reviews focus on modification strategies for enhancing enzyme thermostability, there is a dearth of literature reviewing HTS methods specifically aimed at this objective. Herein, we present a comprehensive overview of various HTS methods utilized for modifying enzyme thermostability across different screening platforms. Additionally, we highlight significant recent examples that demonstrate the successful application of these methods. Furthermore, we address the technical challenges associated with HTS technologies used for screening thermostable enzyme variants and discuss valuable perspectives to promote further advancements in this field. This review serves as an authoritative reference source offering theoretical support for selecting appropriate screening strategies tailored to specific enzymes with the aim of improving their thermostability.

Biosynthesis, acquisition, regulation, and upcycling of heme: recent advances

Heme, an iron-containing tetrapyrrole in hemoproteins, including: hemoglobin, myoglobin, catalase, cytochrome c, and cytochrome P450, plays critical physiological roles in different organisms. Heme-derived chemicals, such as biliverdin, bilirubin, and phycocyanobilin, are known for their antioxidant and anti-inflammatory properties and have shown great potential in fighting viruses and diseases. Therefore, more and more attention has been paid to the biosynthesis of hemoproteins and heme derivatives, which depends on the adequate heme supply in various microbial cell factories. The enhancement of endogenous biosynthesis and exogenous uptake can improve the intracellular heme supply, but the excess free heme is toxic to the cells. Therefore, based on the heme-responsive regulators, several sensitive biosensors were developed to fine-tune the intracellular levels of heme. In this review, recent advances in the: biosynthesis, acquisition, regulation, and upcycling of heme were summarized to provide a solid foundation for the efficient production and application of high-value-added hemoproteins and heme derivatives.

Biosensor-guided discovery and engineering of metabolic enzymes

A variety of chemicals have been produced through metabolic engineering approaches, and enhancing biosynthesis performance can be achieved by using enzymes with high catalytic efficiency. Accordingly, a number of efforts have been made to discover enzymes in nature for various applications. In addition, enzyme engineering approaches have been attempted to suit specific industrial purposes. However, a significant challenge in enzyme discovery and engineering is the efficient screening of enzymes with the desired phenotype from extensive enzyme libraries. To overcome this bottleneck, genetically encoded biosensors have been developed to specifically detect target molecules produced by enzyme activity at the intracellular level. Especially, the biosensors facilitate high-throughput screening (HTS) of targeted enzymes, expanding enzyme discovery and engineering strategies with advances in systems and synthetic biology. This review examines biosensor-guided HTS systems and highlights studies that have utilized these tools to discover enzymes in diverse areas and engineer enzymes to enhance their properties, such as catalytic efficiency, specificity, and stability.

Machine Learning-Guided Protein Engineering

Recent progress in engineering highly promising biocatalysts has increasingly involved machine learning methods. These methods leverage existing experimental and simulation data to aid in the discovery and annotation of promising enzymes, as well as in suggesting beneficial mutations for improving known targets. The field of machine learning for protein engineering is gathering steam, driven by recent success stories and notable progress in other areas. It already encompasses ambitious tasks such as understanding and predicting protein structure and function, catalytic efficiency, enantioselectivity, protein dynamics, stability, solubility, aggregation, and more. Nonetheless, the field is still evolving, with many challenges to overcome and questions to address. In this Perspective, we provide an overview of ongoing trends in this domain, highlight recent case studies, and examine the current limitations of machine learning-based methods. We emphasize the crucial importance of thorough experimental validation of emerging models before their use for rational protein design. We present our opinions on the fundamental problems and outline the potential directions for future research.

Grasping cryptic binding sites to neutralize drug resistance in the field of anticancer

Drug resistance is a significant obstacle to successful cancer treatment. The utilization and development of cryptic binding sites (CBSs) in proteins involved in cancer-related drug-resistance (CRDR) could help to overcome that drug resistance. However, there is no comprehensive review of the successful use of CBSs in addressing CRDR. Here, we have systematically summarized and analyzed the opportunities and challenges of using CBSs in addressing CRDR and revealed the key role that CBSs have in targeting CRDR. First, we have identified the CRDR targets and the corresponding CBSs. Second, we discuss the mechanisms by which CBSs can overcome CRDR. Finally, we have provided examples of successful CBS applications in addressing CRDR. We hope that this approach will provide guidance to biologists and chemists in effectively utilizing CBSs for the development of new drugs to alleviate CRDR.

Directed evolution of Rhodotorula gracilisd-amino acid oxidase using single-cell hydrogel encapsulation and ultrahigh-throughput screening

Engineering catalytic and biophysical properties of enzymes is an essential step en route to advanced biomedical and industrial applications. Here, we developed a high-throughput screening and directed evolution strategy relying on single-cell hydrogel encapsulation to enhance the performance of d-Amino acid oxidase from Rhodotorula gracilis (RgDAAOx), a candidate enzyme for cancer therapy. We used a cascade reaction between RgDAAOx variants surface displayed on yeast and horseradish peroxidase (HRP) in the bulk media to trigger enzyme-mediated crosslinking of phenol-bearing fluorescent alginate macromonomers, resulting in hydrogel formation around single yeast cells. The fluorescent hydrogel capsules served as an artificial phenotype and basis for pooled library screening by fluorescence activated cell sorting (FACS). We screened a RgDAAOx variant library containing ∼10⁶ clones while lowering the d-Ala substrate concentration over three sorting rounds in order to isolate variants with low K_m. After three rounds of FACS sorting and regrowth, we isolated and fully characterized four variants displayed on the yeast surface. We identified variants with a more than 5-fold lower K_m than the parent sequence, with an apparent increase in substrate binding affinity. The mutations we identified were scattered across the RgDAAOx structure, demonstrating the difficulty in rationally predicting allosteric sites and highlighting the advantages of scalable library screening technologies for evolving catalytic enzymes.

Design to Data for mutants of B-glucosidase B from Paenibacillus polymyxa : V311D, F248N, Y166H, Y166K, M221K

Engaging computational tools for protein design is gaining traction in the enzyme engineering community. However, current design and modeling algorithms have limited functionality predictive capacities for enzymes due to limitations of the dataset in terms of size and data quality. This study aims to expand training datasets for improved algorithm development with the addition of five rationally designed single-point enzyme variants. β-glucosidase B variants were modeled in Foldit Standalone and then produced and assayed for thermal stability and kinetic parameters. Functional parameters: thermal stability (T _M ) and Michaelis-Menten constants ( k _cat , K _M , and k _cat /K _M ) of five variants, V311D, Y166H, M221K, F248N, and Y166K, were added into the Design2Data database. As a case study, evaluation of this small mutant set finds mutational effect trends that both corroborate and contradict findings from larger studies examining the entire dataset.

Enzyme Cascade with Horseradish Peroxidase Readout for High-Throughput Screening and Engineering of Human Arginase-1

We report an enzyme cascade with horseradish peroxidase-based readout for screening human arginase-1 (hArg1) activity. We combined the four enzymes hArg1, ornithine decarboxylase, putrescine oxidase, and horseradish peroxidase in a reaction cascade that generated colorimetric or fluorescent signals in response to hArg1 activity and used this cascade to assay wild-type and variant hArg1 sequences as soluble enzymes and displayed on the surface of Escherichia coli. We screened a curated 13-member hArg1 library covering mutations that modified the electrostatic environment surrounding catalytic residues D128 and H141, and identified the R21E variant with a 13% enhanced catalytic turnover rate compared to wild type. Our scalable one-pot single-step arginase assay with continuous kinetic readout is amenable to high-throughput screening and directed evolution of arginase libraries and testing drug candidates for arginase inhibition.

Data-driven strategies for the computational design of enzyme thermal stability: trends, perspectives, and prospects

Thermal stability is one of the most important properties of enzymes, which sustains life and determines the potential for the industrial application of biocatalysts. Although traditional methods such as directed evolution and classical rational design contribute greatly to this field, the enormous sequence space of proteins implies costly and arduous experiments. The development of enzyme engineering focuses on automated and efficient strategies because of the breakthrough of high-throughput DNA sequencing and machine learning models. In this review, we propose a data-driven architecture for enzyme thermostability engineering and summarize some widely adopted datasets, as well as machine learning-driven approaches for designing the thermal stability of enzymes. In addition, we present a series of existing challenges while applying machine learning in enzyme thermostability design, such as the data dilemma, model training, and use of the proposed models. Additionally, a few promising directions for enhancing the performance of the models are discussed. We anticipate that the efficient incorporation of machine learning can provide more insights and solutions for the design of enzyme thermostability in the coming years.

The Synergy between Deep Learning and Organs-on-Chips for High-Throughput Drug Screening: A Review

Organs-on-chips (OoCs) are miniature microfluidic systems that have arguably become a class of advanced in vitro models. Deep learning, as an emerging topic in machine learning, has the ability to extract a hidden statistical relationship from the input data. Recently, these two areas have become integrated to achieve synergy for accelerating drug screening. This review provides a brief description of the basic concepts of deep learning used in OoCs and exemplifies the successful use cases for different types of OoCs. These microfluidic chips are of potential to be assembled as highly potent human-on-chips with complex physiological or pathological functions. Finally, we discuss the future supply with perspectives and potential challenges in terms of combining OoCs and deep learning for image processing and automation designs.

Engineering the Substrate Specificity of Toluene Degrading Enzyme XylM Using Biosensor XylS and Machine Learning

Enzyme engineering using machine learning has been developed in recent years. However, to obtain a large amount of data on enzyme activities for training data, it is necessary to develop a high-throughput and accurate method for evaluating enzyme activities. Here, we examined whether a biosensor-based enzyme engineering method can be applied to machine learning. As a model experiment, we aimed to modify the substrate specificity of XylM, a rate-determining enzyme in a multistep oxidation reaction catalyzed by XylMABC in Pseudomonas putida. XylMABC naturally converts toluene and xylene to benzoic acid and toluic acid, respectively. We aimed to engineer XylM to improve its conversion efficiency to a non-native substrate, 2,6-xylenol. Wild-type XylMABC slightly converted 2,6-xylenol to 3-methylsalicylic acid, which is the ligand of the transcriptional regulator XylS in P. putida. By locating a fluorescent protein gene under the control of the Pm promoter to which XylS binds, a XylS-producing Escherichia coli strain showed higher fluorescence intensity in a 3-methylsalicylic acid concentration-dependent manner. We evaluated the 3-methylsalicylic acid productivity of XylM variants using the fluorescence intensity of the sensor strain as an indicator. The obtained data provided the training data for machine learning for the directed evolution of XylM. Two cycles of machine learning-assisted directed evolution resulted in the acquisition of XylM-D140E-V144K-F243L-N244S with 15 times higher productivity than wild-type XylM. These results demonstrate that an indirect enzyme activity evaluation method using biosensors is sufficiently quantitative and high-throughput to be used as training data for machine learning. The findings expand the versatility of machine learning in enzyme engineering.