Machine Learning in Enzyme Engineering

Bacterial α-Glucan and Branching Sucrases from GH70 Family: Discovery, Structure-Function Relationship Studies and Engineering

Glucansucrases and branching sucrases are classified in the family 70 of glycoside hydrolases. They are produced by lactic acid bacteria occupying very diverse ecological niches (soil, buccal cavity, sourdough, intestine, dairy products, etc.). Usually secreted by their producer organisms, they are involved in the synthesis of α-glucans from sucrose substrate. They contribute to cell protection while promoting adhesion and colonization of different biotopes. Dextran, an α-1,6 linked linear α-glucan, was the first microbial polysaccharide commercialized for medical applications. Advances in the discovery and characterization of these enzymes have remarkably enriched the available diversity with new catalysts. Research into their molecular mechanisms has highlighted important features governing their peculiarities thus opening up many opportunities for engineering these catalysts to provide new routes for the transformation of sucrose into value-added molecules. This article reviews these different aspects with the ambition to show how they constitute the basis for promising future developments.

Recent trends in biocatalysis

Biocatalysis has undergone revolutionary progress in the past century. Benefited by the integration of multidisciplinary technologies, natural enzymatic reactions are constantly being explored. Protein engineering gives birth to robust biocatalysts that are widely used in industrial production. These research achievements have gradually constructed a network containing natural enzymatic synthesis pathways and artificially designed enzymatic cascades. Nowadays, the development of artificial intelligence, automation, and ultra-high-throughput technology provides infinite possibilities for the discovery of novel enzymes, enzymatic mechanisms and enzymatic cascades, and gradually complements the lack of remaining key steps in the pathway design of enzymatic total synthesis. Therefore, the research of biocatalysis is gradually moving towards the era of novel technology integration, intelligent manufacturing and enzymatic total synthesis.

Methods for enzyme library creation: Which one will you choose?: A guide for novices and experts to introduce genetic diversity

Enzyme engineering allows to explore sequence diversity in search for new properties. The scientific literature is populated with methods to create enzyme libraries for engineering purposes, however, choosing a suitable method for the creation of mutant libraries can be daunting, in particular for the novices. Here, we address both novices and experts: how can one enter the arena of enzyme library design and what guidelines can advanced users apply to select strategies best suited to their purpose? Section I is dedicated to the novices and presents an overview of established and standard methods for library creation, as well as available commercial solutions. The expert will discover an up-to-date tool to freshen up their repertoire (Section I) and learn of the newest methods that are likely to become a mainstay (Section II). We focus primarily on in vitro methods, presenting the advantages of each method. Our ultimate aim is to offer a selection of methods/strategies that we believe to be most useful to the enzyme engineer, whether a first-timer or a seasoned user.

Structural model and functional properties of an exo-polygalacturonase from Neosartorya glabra

A purified exo-polygalacturonase of Neosartorya glabra (EplNg) was successfully characterized. EplNg native presented 68.2 kDa, with 32% carbohydrate content. The deglycosylated form showed 46.3 kDa and isoelectric point of 5.4. The identity of EplNg was confirmed as an exo-polygalacturonase class I (EC 3.2.1.67) using mass spectrometry and Western-Blotting. Capillary electrophoresis indicated that only galacturonic acid was released by the action of EplNg on sodium polypectate, confirming an exoenzyme character. The structural model confers that EplNg has a core formed by twisted parallel β-sheets structure. Among twelve putative cysteines, ten were predicted to form disulfide bridges. The catalytic triad predicted is composed of Asp²²³, Asp²⁴⁵, and Asp²⁴⁶ aligned along with a distance in 4-5 Å, suggesting that EplNg probably does not perform the standard inverting catalytic mechanism described for the GH28 family. EplNg was active from 30 to 90 °C, with maximum activity at 65 °C, pH 5.0. The Km and Vmax determined using sodium polypectate were 6.9 mg·mL^-1 and Vmax 690 μmol·min^-1.mg^-1, respectively. EplNg was active and stable over a wide range of pH values and temperatures, confirming the interesting properties EplNg and provide a basis for the development of the enzyme in different biotechnological processes.

PyPEF-An Integrated Framework for Data-Driven Protein Engineering

Data-driven strategies are gaining increased attention in protein engineering due to recent advances in access to large experimental databanks of proteins, next-generation sequencing (NGS), high-throughput screening (HTS) methods, and the development of artificial intelligence algorithms. However, the reliable prediction of beneficial amino acid substitutions, their combination, and the effect on functional properties remain the most significant challenges in protein engineering, which is applied to develop proteins and enzymes for biocatalysis, biomedicine, and life sciences. Here, we present a general-purpose framework (PyPEF: pythonic protein engineering framework) for performing data-driven protein engineering using machine learning methods combined with techniques from signal processing and statistical physics. PyPEF guides the identification and selection of beneficial proteins of a defined sequence space by systematically or randomly exploring the fitness of variants and by sampling random evolution pathways. The performance of PyPEF was evaluated concerning its predictive accuracy and throughput on four public protein and enzyme data sets using common regression models. It was proved that the program could efficiently predict the fitness of protein sequences for different target properties (predictive models with coefficient of determination values ranging from 0.58 to 0.92). By combining machine learning and protein evolution, PyPEF enabled the screening of proteins with various functions, reaching a screening capacity of more than 500,000 protein sequence variants in the timeframe of only a few minutes on a personal computer. PyPEF displayed significant accuracies on four public data sets (different proteins and properties) and underlined the potential of integrating data-driven technologies for covering different philosophies by either predicting the fitness of the variants to the highest accuracy accounting for epistatic effects or capturing the general trend of introduced mutations on the fitness in directed protein evolution campaigns. In essence, PyPEF can provide a powerful solution to current sequence exploration and combinatorial problems faced in protein engineering through exhaustive in silico screening of the sequence space.

Computer-aided understanding and engineering of enzymatic selectivity

Enzymes offering chemo-, regio-, and stereoselectivity enable the asymmetric synthesis of high-value chiral molecules. Unfortunately, the drawback that naturally occurring enzymes are often inefficient or have undesired selectivity toward non-native substrates hinders the broadening of biocatalytic applications. To match the demands of specific selectivity in asymmetric synthesis, biochemists have implemented various computer-aided strategies in understanding and engineering enzymatic selectivity, diversifying the available repository of artificial enzymes. Here, given that the entire asymmetric catalytic cycle, involving precise interactions within the active pocket and substrate transport in the enzyme channel, could affect the enzymatic efficiency and selectivity, we presented a comprehensive overview of the computer-aided workflow for enzymatic selectivity. This review includes a mechanistic understanding of enzymatic selectivity based on quantum mechanical calculations, rational design of enzymatic selectivity guided by enzyme-substrate interactions, and enzymatic selectivity regulation via enzyme channel engineering. Finally, we discussed the computational paradigm for designing enzyme selectivity in silico to facilitate the advancement of asymmetric biosynthesis.

Structural-Based Modeling in Protein Engineering. A Must Do

Biotechnological solutions will be a key aspect in our immediate future society, where optimized enzymatic processes through enzyme engineering might be an important solution for waste transformation, clean energy production, biodegradable materials, and green chemistry, for example. Here we advocate the importance of structural-based bioinformatics and molecular modeling tools in such developments. We summarize our recent experiences indicating a great prediction/success ratio, and we suggest that an early in silico phase should be performed in enzyme engineering studies. Moreover, we demonstrate the potential of a new technique combining Rosetta and PELE, which could provide a faster and more automated procedure, an essential aspect for a broader use.

Engineering the protein dynamics of an ancestral luciferase

Protein dynamics are often invoked in explanations of enzyme catalysis, but their design has proven elusive. Here we track the role of dynamics in evolution, starting from the evolvable and thermostable ancestral protein AncHLD-RLuc which catalyses both dehalogenase and luciferase reactions. Insertion-deletion (InDel) backbone mutagenesis of AncHLD-RLuc challenged the scaffold dynamics. Screening for both activities reveals InDel mutations localized in three distinct regions that lead to altered protein dynamics (based on crystallographic B-factors, hydrogen exchange, and molecular dynamics simulations). An anisotropic network model highlights the importance of the conformational flexibility of a loop-helix fragment of Renilla luciferases for ligand binding. Transplantation of this dynamic fragment leads to lower product inhibition and highly stable glow-type bioluminescence. The success of our approach suggests that a strategy comprising (i) constructing a stable and evolvable template, (ii) mapping functional regions by backbone mutagenesis, and (iii) transplantation of dynamic features, can lead to functionally innovative proteins.

How Deep Learning Tools Can Help Protein Engineers Find Good Sequences

The deep learning revolution introduced a new and efficacious way to address computational challenges in a wide range of fields, relying on large data sets and powerful computational resources. In protein engineering, we consider the challenge of computationally predicting properties of a protein and designing sequences with these properties. Indeed, accurate and fast deep network oracles for different properties of proteins have been developed. These learn to predict a property from an amino acid sequence by training on large sets of proteins that have this property. In particular, deep networks can learn from the set of all known protein sequences to identify ones that are protein-like. A fundamental challenge when engineering sequences that are both protein-like and satisfy a desired property is that these are rare instances within the vast space of all possible ones. When searching for these very rare instances, one would like to use good sampling procedures. Sampling approaches that are decoupled from the prediction of the property or in which the predictor uses only post-sampling to identify good instances are less efficient. The alternative is to use sampling methods that are geared to generate sequences satisfying and/or optimizing the predictor's desired properties. Deep learning has a class of architectures, denoted as generative models, which offer the capability of sampling from the learned distribution of a predicted property. Here, we review the use of deep learning tools to find good sequences for protein engineering, including developing oracles/predictors of a property of the proteins and methods that sample from a distribution of protein-like sequences to optimize the desired property.

Enhancement of protein thermostability by three consecutive mutations using loop-walking method and machine learning

We developed a method to improve protein thermostability, “loop-walking method”. Three consecutive positions in 12 loops of Burkholderia cepacia lipase were subjected to random mutagenesis to make 12 libraries. Screening allowed us to identify L7 as a hot-spot loop having an impact on thermostability, and the P233G/L234E/V235M mutant was found from 214 variants in the L7 library. Although a more excellent mutant might be discovered by screening all the 8000 P233X/L234X/V235X mutants, it was difficult to assay all of them. We therefore employed machine learning. Using thermostability data of the 214 mutants, a computational discrimination model was constructed to predict thermostability potentials. Among 7786 combinations ranked in silico, 20 promising candidates were selected and assayed. The P233D/L234P/V235S mutant retained 66% activity after heat treatment at 60 °C for 30 min, which was higher than those of the wild-type enzyme (5%) and the P233G/L234E/V235M mutant (35%).

Protein sequence design with deep generative models

Protein engineering seeks to identify protein sequences with optimized properties. When guided by machine learning, protein sequence generation methods can draw on prior knowledge and experimental efforts to improve this process. In this review, we highlight recent applications of machine learning to generate protein sequences, focusing on the emerging field of deep generative methods.

Recent advances in user-friendly computational tools to engineer protein function

Progress in technology and algorithms throughout the past decade has transformed the field of protein design and engineering. Computational approaches have become well-engrained in the processes of tailoring proteins for various biotechnological applications. Many tools and methods are developed and upgraded each year to satisfy the increasing demands and challenges of protein engineering. To help protein engineers and bioinformaticians navigate this emerging wave of dedicated software, we have critically evaluated recent additions to the toolbox regarding their application for semi-rational and rational protein engineering. These newly developed tools identify and prioritize hotspots and analyze the effects of mutations for a variety of properties, comprising ligand binding, protein-protein and protein-nucleic acid interactions, and electrostatic potential. We also discuss notable progress to target elusive protein dynamics and associated properties like ligand-transport processes and allosteric communication. Finally, we discuss several challenges these tools face and provide our perspectives on the further development of readily applicable methods to guide protein engineering efforts.

Revolutionizing enzyme engineering through artificial intelligence and machine learning

The combinatorial space of an enzyme sequence has astronomical possibilities and exploring it with contemporary experimental techniques is arduous and often ineffective. Multi-target objectives such as concomitantly achieving improved selectivity, solubility and activity of an enzyme have narrow plausibility under approaches of restricted mutagenesis and combinatorial search. Traditional enzyme engineering approaches have a limited scope for complex optimization due to the requirement of a priori knowledge or experimental burden of screening huge protein libraries. The recent surge in high-throughput experimental methods including Next Generation Sequencing and automated screening has flooded the field of molecular biology with big-data, which requires us to re-think our concurrent approaches towards enzyme engineering. Artificial Intelligence (AI) and Machine Learning (ML) have great potential to revolutionize smart enzyme engineering without the explicit need for a complete understanding of the underlying molecular system. Here, we portray the role and position of AI techniques in the field of enzyme engineering along with their scope and limitations. In addition, we explain how the traditional approaches of directed evolution and rational design can be extended through AI tools. Recent successful examples of AI-assisted enzyme engineering projects and their deviation from traditional approaches are highlighted. A comprehensive picture of current challenges and future avenues for AI in enzyme engineering are also discussed.

Robotics for enzyme technology: innovations and technological perspectives

The use of robotics in the life science sector has created a considerable and significant impact on a wide range of research areas, including enzyme technology due to their immense applications in enzyme and microbial engineering as an indispensable tool in high-throughput screening applications. Scientists are experiencing the advanced applications of various biological robots (nanobots), fabricated based on bottom-up or top-down approaches for making nanotechnology scaffolds. Nanobots and enzyme-powered nanomotors are particularly attractive because they are self-propelled vehicles, which consume biocompatible fuels. These smart nanostructures are widely used as drug delivery systems for the efficient treatment of various diseases. This review gives insights into the escalating necessity of robotics and nanobots and their ever-widening applications in enzyme technology, including biofuel production and biomedical applications. It also offers brief insights into high-throughput robotic platforms that are currently being used in enzyme screening applications for monitoring and control of microbial growth conditions. KEY POINTS: • Robotics and their applications in biotechnology are highlighted. • Robotics for high-throughput enzyme screening and microbial engineering are described. • Nanobots and enzyme-powered nanomotors as controllable drug delivery systems are reviewed.

Machine learning classifiers aid virtual screening for efficient design of mini-protein therapeutics

De novo design of mini-proteins (4-12 kDa) has recently been shown to produce new candidates for protein therapeutics. They are temperature stable molecules that bind to the drug target with high affinity for inhibiting its interactions. The development of mini-protein binders requires laboratory screening of tens of thousands of molecules for effective target binding. In this study we trained machine learning classifiers which can distinguish, with 90% accuracy and 80% precision, mini-protein binders from non-binding molecules designed for a particular target; this significantly reduces the number of mini protein candidates for experimental screening. Further, on the basis of our results we propose a multi-stage protocol where a small dataset (few hundred experimentally verified target-specific mini-proteins) can be used to train classifiers for improving the efficiency of mini-protein design for any specific target.

Three Simple Properties Explain Protein Stability Change upon Mutation

Accurate prediction of protein stability upon mutation enables rational engineering of new proteins and insights into protein evolution and monogenetic diseases caused by single-point amino acid substitutions. Many tools have been developed to this aim, ranging from energy-based models to machine-learning methods that use large amounts of experimental data. However, as the methods become more complex, the interpretation of the chemistry underlying the protein stability effects becomes obscure. It is thus of interest to identify the simplest prediction model that retains complete amino acid specific interpretation; for a given number of input descriptors, we expect such a model to be almost universal. In this study, we identify such a limiting model, SimBa, a simple multilinear regression model trained on a substitution-type-balanced experimental data set. The model accounts only for the solvent accessibility of the site, volume difference, and polarity difference caused by mutation. Our results show that this very simple and directly applicable model performs comparably to other much more complex, widely used protein stability prediction methods. This suggests that a hard limit of ∼1 kcal/mol numerical accuracy and an R ∼ 0.5 trend accuracy exists and that new features, such as account of unfolded states, water colocalization, and amino acid correlations, are required to improve accuracy to, e.g., 1/2 kcal/mol.

100th Anniversary of Macromolecular Science Viewpoint: Data-Driven Protein Design

The design of synthetic proteins with the desired function is a long-standing goal in biomolecular science, with broad applications in biochemical engineering, agriculture, medicine, and public health. Rational de novo design and experimental directed evolution have achieved remarkable successes but are challenged by the requirement to find functional "needles" in the vast "haystack" of protein sequence space. Data-driven models for fitness landscapes provide a predictive map between protein sequence and function and can prospectively identify functional candidates for experimental testing to greatly improve the efficiency of this search. This Viewpoint reviews the applications of machine learning and, in particular, deep learning as part of data-driven protein engineering platforms. We highlight recent successes, review promising computational methodologies, and provide an outlook on future challenges and opportunities. The article is written for a broad audience comprising both polymer and protein scientists and computer and data scientists interested in an up-to-date review of recent innovations and opportunities in this rapidly evolving field.

Enzyme Discovery and Engineering for Sustainable Plastic Recycling

The drastically increasing amount of plastic waste is causing an environmental crisis that requires innovative technologies for recycling post-consumer plastics to achieve waste valorization while meeting environmental quality goals. Biocatalytic depolymerization mediated by enzymes has emerged as an efficient and sustainable alternative for plastic treatment and recycling. A variety of plastic-degrading enzymes have been discovered from microbial sources. Meanwhile, protein engineering has been exploited to modify and optimize plastic-degrading enzymes. This review highlights the recent trends and up-to-date advances in mining novel plastic-degrading enzymes through state-of-the-art omics-based techniques and improving the enzyme catalytic efficiency and stability via various protein engineering strategies. Future research prospects and challenges are also discussed.

Machine Learning Enables Selection of Epistatic Enzyme Mutants for Stability Against Unfolding and Detrimental Aggregation

Machine learning (ML) has pervaded most areas of protein engineering, including stability and stereoselectivity. Using limonene epoxide hydrolase as the model enzyme and innov'SAR as the ML platform, comprising a digital signal process, we achieved high protein robustness that can resist unfolding with concomitant detrimental aggregation. Fourier transform (FT) allows us to take into account the order of the protein sequence and the nonlinear interactions between positions, and thus to grasp epistatic phenomena. The innov'SAR approach is interpolative, extrapolative and makes outside-the-box, predictions not found in other state-of-the-art ML or deep learning approaches. Equally significant is the finding that our approach to ML in the present context, flanked by advanced molecular dynamics simulations, uncovers the connection between epistatic mutational interactions and protein robustness.

Recent advances in artificial enzyme cascades for the production of value-added chemicals

Enzyme cascades are efficient tools to perform multi-step synthesis in one-pot in a green and sustainable manner, enabling non-natural synthesis of valuable chemicals from easily available substrates by artificially combining two or more enzymes. Bioproduction of many high-value chemicals such as chiral and highly functionalised molecules have been achieved by developing new enzyme cascades. This review summarizes recent advances on engineering and application of enzyme cascades to produce high-value chemicals (alcohols, aldehydes, ketones, amines, carboxylic acids, etc) from simple starting materials. While 2-step enzyme cascades are developed for versatile enantioselective synthesis, multi-step enzyme cascades are engineered to functionalise basic chemicals, such as styrenes, cyclic alkanes, and aromatic compounds. New cascade reactions have also been developed for producing valuable chemicals from bio-based substrates, such as ʟ-phenylalanine, and renewable feedstocks such as glucose and glycerol. The challenges in current process and future outlooks in the development of enzyme cascades are also addressed.

Advances in machine learning for directed evolution

Machine learning (ML) can expedite directed evolution by allowing researchers to move expensive experimental screens in silico. Gathering sequence-function data for training ML models, however, can still be costly. In contrast, raw protein sequence data is widely available. Recent advances in ML approaches use protein sequences to augment limited sequence-function data for directed evolution. We highlight contributions in a growing effort to use sequences to reduce or eliminate the amount of sequence-function data needed for effective in silico screening. We also highlight approaches that use ML models trained on sequences to generate new functional sequence diversity, focusing on strategies that use these generative models to efficiently explore vast regions of protein space.

Enzyme engineering and its industrial applications

Recently, there has been an increase in the demand for enzymes with modified activity, specificity, and stability. Enzyme engineering is an important tool to meet the demand for enzymes adjusted to different industrial processes. Knowledge of the structure and function of enzymes guides the choice of the best strategy for engineering enzymes. Each enzyme engineering strategy, such as rational design, directed evolution, and semi-rational design, has specific applications, as well as limitations, which must be considered when choosing a suitable strategy. Engineered enzymes can be optimized for different industrial applications by choosing the appropriate strategy. This review features engineered enzymes that have been applied in food, animal feed, pharmaceuticals, medical applications, bioremediation, biofuels, and detergents.

Ionic liquids for regulating biocatalytic process: Achievements and perspectives

Biocatalysis has found enormous applications in sorts of fields as an alternative to chemical catalysis. In the pursue of green and sustainable chemistry, ionic liquids (ILs) have been considered as promising reaction media for biocatalysis, owing to their unique characteristics, such as nonvolatility, inflammability and tunable properties as regards polarity and water miscibility behavior, compared to organic solvents. In recent years, great developments have been achieved in respects to biocatalysis in ILs, especially for preparing various chemicals. This review tends to give illustrative examples with a focus on representative chemicals production by biocatalyst in ILs and elucidate the possible mechanism in such systems. It also discusses how to regulate the catalytic efficiency from several aspects and finally provides an outlook on the opportunities to broaden biocatalysis in ILs.

Power of Biocatalysis for Organic Synthesis

Biocatalysis, using defined enzymes for organic transformations, has become a common tool in organic synthesis, which is also frequently applied in industry. The generally high activity and outstanding stereo-, regio-, and chemoselectivity observed in many biotransformations are the result of a precise control of the reaction in the active site of the biocatalyst. This control is achieved by exact positioning of the reagents relative to each other in a fine-tuned 3D environment, by specific activating interactions between reagents and the protein, and by subtle movements of the catalyst. Enzyme engineering enables one to adapt the catalyst to the desired reaction and process. A well-filled biocatalytic toolbox is ready to be used for various reactions. Providing nonnatural reagents and conditions and evolving biocatalysts enables one to play with the myriad of options for creating novel transformations and thereby opening new, short pathways to desired target molecules. Combining several biocatalysts in one pot to perform several reactions concurrently increases the efficiency of biocatalysis even further.

Computational design of enzymes for biotechnological applications

Enzymes are the natural catalysts that execute biochemical reactions upholding life. Their natural effectiveness has been fine-tuned as a result of millions of years of natural evolution. Such catalytic effectiveness has prompted the use of biocatalysts from multiple sources on different applications, including the industrial production of goods (food and beverages, detergents, textile, and pharmaceutics), environmental protection, and biomedical applications. Natural enzymes often need to be improved by protein engineering to optimize their function in non-native environments. Recent technological advances have greatly facilitated this process by providing the experimental approaches of directed evolution or by enabling computer-assisted applications. Directed evolution mimics the natural selection process in a highly accelerated fashion at the expense of arduous laboratory work and economic resources. Theoretical methods provide predictions and represent an attractive complement to such experiments by waiving their inherent costs. Computational techniques can be used to engineer enzymatic reactivity, substrate specificity and ligand binding, access pathways and ligand transport, and global properties like protein stability, solubility, and flexibility. Theoretical approaches can also identify hotspots on the protein sequence for mutagenesis and predict suitable alternatives for selected positions with expected outcomes. This review covers the latest advances in computational methods for enzyme engineering and presents many successful case studies.

Biocatalytic routes to anti-viral agents and their synthetic intermediates

An assessment of biocatalytic strategies for the synthesis of anti-viral agents, offering guidelines for the development of sustainable production methods for a future COVID-19 remedy.

Flow biocatalysis 101: design, development and applications

Flow biocatalysis: where to start? This tutorial review aims to guide and inspire new-comers to the field to boost the potential of flow biocatalysis.

Looking Back: A Short History of the Discovery of Enzymes and How They Became Powerful Chemical Tools

Enzymatic approaches to challenges in chemical synthesis are increasingly popular and very attractive to industry given their green nature and high efficiency compared to traditional methods. In this historical review we highlight the developments across several fields that were necessary to create the modern field of biocatalysis, with enzyme engineering and directed evolution at its core. We exemplify the modular, incremental, and highly unpredictable nature of scientific discovery, driven by curiosity, and showcase the resulting examples of cutting-edge enzymatic applications in industry.

Synthetic biology 2020-2030: six commercially-available products that are changing our world

Synthetic biology will transform how we grow food, what we eat, and where we source materials and medicines. Here I have selected six products that are now on the market, highlighting the underlying technologies and projecting forward to the future that can be expected over the next ten years.

UMI-linked consensus sequencing enables phylogenetic analysis of directed evolution

The success of protein evolution campaigns is strongly dependent on the sequence context in which mutations are introduced, stemming from pervasive non-additive interactions between a protein's amino acids ('intra-gene epistasis'). Our limited understanding of such epistasis hinders the correct prediction of the functional contributions and adaptive potential of mutations. Here we present a straightforward unique molecular identifier (UMI)-linked consensus sequencing workflow (UMIC-seq) that simplifies mapping of evolutionary trajectories based on full-length sequences. Attaching UMIs to gene variants allows accurate consensus generation for closely related genes with nanopore sequencing. We exemplify the utility of this approach by reconstructing the artificial phylogeny emerging in three rounds of directed evolution of an amine dehydrogenase biocatalyst via ultrahigh throughput droplet screening. Uniquely, we are able to identify lineages and their founding variant, as well as non-additive interactions between mutations within a full gene showing sign epistasis. Access to deep and accurate long reads will facilitate prediction of key beneficial mutations and adaptive potential based on in silico analysis of large sequence datasets.

Inferring Protein Sequence-Function Relationships with Large-Scale Positive-Unlabeled Learning

Machine learning can infer how protein sequence maps to function without requiring a detailed understanding of the underlying physical or biological mechanisms. It is challenging to apply existing supervised learning frameworks to large-scale experimental data generated by deep mutational scanning (DMS) and related methods. DMS data often contain high-dimensional and correlated sequence variables, experimental sampling error and bias, and the presence of missing data. Notably, most DMS data do not contain examples of negative sequences, making it challenging to directly estimate how sequence affects function. Here, we develop a positive-unlabeled (PU) learning framework to infer sequence-function relationships from large-scale DMS data. Our PU learning method displays excellent predictive performance across ten large-scale sequence-function datasets, representing proteins of different folds, functions, and library types. The estimated parameters pinpoint key residues that dictate protein structure and function. Finally, we apply our statistical sequence-function model to design highly stabilized enzymes.

MCIC: Automated Identification of Cellulases From Metagenomic Data and Characterization Based on Temperature and pH Dependence

As the availability of high-throughput metagenomic data is increasing, agile and accurate tools are required to analyze and exploit this valuable and plentiful resource. Cellulose-degrading enzymes have various applications, and finding appropriate cellulases for different purposes is becoming increasingly challenging. An in silico screening method for high-throughput data can be of great assistance when combined with the characterization of thermal and pH dependence. By this means, various metagenomic sources with high cellulolytic potentials can be explored. Using a sequence similarity-based annotation and an ensemble of supervised learning algorithms, this study aims to identify and characterize cellulolytic enzymes from a given high-throughput metagenomic data based on optimum temperature and pH. The prediction performance of MCIC (metagenome cellulase identification and characterization) was evaluated through multiple iterations of sixfold cross-validation tests. This tool was also implemented for a comparative analysis of four metagenomic sources to estimate their cellulolytic profile and capabilities. For experimental validation of MCIC’s screening and prediction abilities, two identified enzymes from cattle rumen were subjected to cloning, expression, and characterization. To the best of our knowledge, this is the first time that a sequence-similarity based method is used alongside an ensemble machine learning model to identify and characterize cellulase enzymes from extensive metagenomic data. This study highlights the strength of machine learning techniques to predict enzymatic properties solely based on their sequence. MCIC is freely available as a python package and standalone toolkit for Windows and Linux-based operating systems with several functions to facilitate the screening and thermal and pH dependence prediction of cellulases.

In silico design and automated learning to boost next-generation smart biomanufacturing

The increasing demand for bio-based compounds produced from waste or sustainable sources is driving biofoundries to deliver a new generation of prototyping biomanufacturing platforms. Integration and automation of the design, build, test and learn (DBTL) steps in centers like SYNBIOCHEM in Manchester and across the globe (Global Biofoundries Alliance) are helping to reduce the delivery time from initial strain screening and prototyping towards industrial production. Notably, a portfolio of producer strains for a suite of material monomers was recently developed, some approaching industrial titers, in a tour de force by the Manchester Centre that was achieved in less than 90 days. New in silico design tools are providing significant contributions to the front end of the DBTL pipelines. At the same time, the far-reaching initiatives of modern biofoundries are generating a large amount of high-dimensional data and knowledge that can be integrated through automated learning to expedite the DBTL cycle. In this Perspective, the new design tools and the role of the learning component as an enabling technology for the next generation of automated biofoundries are discussed. Future biofoundries will operate under completely automated DBTL cycles driven by in silico optimal experimental planning, full biomanufacturing devices connectivity, virtualization platforms and cloud-based design. The automated generation of robotic build worklists and the integration of machine-learning algorithms will collectively allow high levels of adaptability and rapid design changes toward fully automated smart biomanufacturing.

Decoding the intricate network of molecular interactions of a hyperstable engineered biocatalyst

Computational design of protein catalysts with enhanced stabilities for use in research and enzyme technologies is a challenging task. Using force-field calculations and phylogenetic analysis, we previously designed the haloalkane dehalogenase DhaA115 which contains 11 mutations that confer upon it outstanding thermostability (T m = 73.5 °C; ΔT m > 23 °C). An understanding of the structural basis of this hyperstabilization is required in order to develop computer algorithms and predictive tools. Here, we report X-ray structures of DhaA115 at 1.55 Å and 1.6 Å resolutions and their molecular dynamics trajectories, which unravel the intricate network of interactions that reinforce the αβα-sandwich architecture. Unexpectedly, mutations toward bulky aromatic amino acids at the protein surface triggered long-distance (∼27 Å) backbone changes due to cooperative effects. These cooperative interactions produced an unprecedented double-lock system that: (i) induced backbone changes, (ii) closed the molecular gates to the active site, (iii) reduced the volumes of the main and slot access tunnels, and (iv) occluded the active site. Despite these spatial restrictions, experimental tracing of the access tunnels using krypton derivative crystals demonstrates that transport of ligands is still effective. Our findings highlight key thermostabilization effects and provide a structural basis for designing new thermostable protein catalysts.

Increasing metabolic pathway flux by using machine learning models

Machine learning is transforming many industries through self-improving models that are fueled by big data and high computing power. The field of metabolic engineering, which uses cellular biochemical network to manufacture useful small molecules, has also witnessed the first wave of machine learning applications in the past five years, covering reaction route design, enzyme selection, pathway engineering and process optimization. This review focuses on pathway engineering, and uses a few recent studies to illustrate (1) how machine learning models can be useful in overcoming an evident rate-limiting step, and (2) how the models may be used to exhaustively search - or guide optimization algorithms to search - a large design space when the cellular regulation of the reaction network is more convoluted.

Ultrahigh throughput screening for enzyme function in droplets

Water-in-oil droplets, made and handled in microfluidic devices, provide a new experimental format, in which ultrahigh throughput experiments can be conducted faster and with minimal reagent consumption. An increasing number of studies have emerged that applied this approach to directed evolution and metagenomic screening of enzyme catalysts. Here, we review the considerations necessary to implement robust workflows, based on choices of device design, detection modes, emulsion formulations and substrates, and scope out which enzyme classes have become amenable to droplet screening.

Harnessing Conformational Plasticity to Generate Designer Enzymes

Recent years have witnessed an explosion of interest in understanding the role of conformational dynamics both in the evolution of new enzymatic activities from existing enzymes and in facilitating the emergence of enzymatic activity de novo on scaffolds that were previously non-catalytic. There are also an increasing number of examples in the literature of targeted engineering of conformational dynamics being successfully used to alter enzyme selectivity and activity. Despite the obvious importance of conformational dynamics to both enzyme function and evolvability, many (although not all) computational design approaches still focus either on pure sequence-based approaches or on using structures with limited flexibility to guide the design. However, there exist a wide variety of computational approaches that can be (re)purposed to introduce conformational dynamics as a key consideration in the design process. Coupled with laboratory evolution and more conventional existing sequence- and structure-based approaches, these techniques provide powerful tools for greatly expanding the protein engineering toolkit. This Perspective provides an overview of evolutionary studies that have dissected the role of conformational dynamics in facilitating the emergence of novel enzymes, as well as advances in computational approaches that allow one to target conformational dynamics as part of enzyme design. Harnessing conformational dynamics in engineering studies is a powerful paradigm with which to engineer the next generation of designer biocatalysts.

Enhancing a de novo enzyme activity by computationally-focused ultra-low-throughput screening

Directed evolution has revolutionized protein engineering. Still, enzyme optimization by random library screening remains sluggish, in large part due to futile probing of mutations that are catalytically neutral and/or impair stability and folding. FuncLib is a novel approach which uses phylogenetic analysis and Rosetta design to rank enzyme variants with multiple mutations, on the basis of predicted stability. Here, we use it to target the active site region of a minimalist-designed, de novo Kemp eliminase. The similarity between the Michaelis complex and transition state for the enzymatic reaction makes this system particularly challenging to optimize. Yet, experimental screening of a small number of active-site variants at the top of the predicted stability ranking leads to catalytic efficiencies and turnover numbers (∼2 × 104 M-1 s-1 and ∼102 s-1) for this anthropogenic reaction that compare favorably to those of modern natural enzymes. This result illustrates the promise of FuncLib as a powerful tool with which to speed up directed evolution, even on scaffolds that were not originally evolved for those functions, by guiding screening to regions of the sequence space that encode stable and catalytically diverse enzymes. Empirical valence bond calculations reproduce the experimental activation energies for the optimized eliminases to within ∼2 kcal mol-1 and indicate that the enhanced activity is linked to better geometric preorganization of the active site. This raises the possibility of further enhancing the stability-guidance of FuncLib by computational predictions of catalytic activity, as a generalized approach for computational enzyme design.

Practically useful protein-design methods combining phylogenetic and atomistic calculations

Our ability to design new or improved biomolecular activities depends on understanding the sequence-function relationships in proteins. The large size and fold complexity of most proteins, however, obscure these relationships, and protein-optimization methods continue to rely on laborious experimental iterations. Recently, a deeper understanding of the roles of stability-threshold effects and biomolecular epistasis in proteins has led to the development of hybrid methods that combine phylogenetic analysis with atomistic design calculations. These methods enable reliable and even single-step optimization of protein stability, expressibility, and activity in proteins that were considered outside the scope of computational design. Furthermore, ancestral-sequence reconstruction produces insights on missing links in the evolution of enzymes and binders that may be used in protein design. Through the combination of phylogenetic and atomistic calculations, the long-standing goal of general computational methods that can be universally applied to study and optimize proteins finally seems within reach.

One Pot Use of Combilipases for Full Modification of Oils and Fats: Multifunctional and Heterogeneous Substrates

Lipases are among the most utilized enzymes in biocatalysis. In many instances, the main reason for their use is their high specificity or selectivity. However, when full modification of a multifunctional and heterogeneous substrate is pursued, enzyme selectivity and specificity become a problem. This is the case of hydrolysis of oils and fats to produce free fatty acids or their alcoholysis to produce biodiesel, which can be considered cascade reactions. In these cases, to the original heterogeneity of the substrate, the presence of intermediate products, such as diglycerides or monoglycerides, can be an additional drawback. Using these heterogeneous substrates, enzyme specificity can promote that some substrates (initial substrates or intermediate products) may not be recognized as such (in the worst case scenario they may be acting as inhibitors) by the enzyme, causing yields and reaction rates to drop. To solve this situation, a mixture of lipases with different specificity, selectivity and differently affected by the reaction conditions can offer much better results than the use of a single lipase exhibiting a very high initial activity or even the best global reaction course. This mixture of lipases from different sources has been called “combilipases” and is becoming increasingly popular. They include the use of liquid lipase formulations or immobilized lipases. In some instances, the lipases have been coimmobilized. Some discussion is offered regarding the problems that this coimmobilization may give rise to, and some strategies to solve some of these problems are proposed. The use of combilipases in the future may be extended to other processes and enzymes.