Accelerating Biocatalysis Discovery with Machine Learning: A Paradigm Shift in Enzyme Engineering, Discovery, and Design

A mini review on revolutionizing hydrogenation catalysis: unleashing transformative power of artificial intelligence

CONTEXT

The field of hydrogenation catalysis has undergone a revolution due to the application of artificial intelligence (AI) and machine learning (ML), which have opened up new avenues for improving catalyst design, reaction efficiency, and pathway optimization. Trial-and-error techniques are a major component of traditional catalyst discovery methods, and they can be resource and time-intensive when it comes to real-world applications. On the other hand, real-time reaction condition optimization, predictive modelling, and quicker catalyst screening are made possible by AI-based techniques. Using methods like neural networks, Bayesian optimization, and generative models, this paper emphasizes how artificial intelligence has been revolutionizing catalyst creation along with mechanistic knowledge and process intensification. AI has the ability to completely transform catalytic research, as demonstrated by a number of case studies that demonstrate its use in CO₂ hydrogenation, biomass upgrading, and metal catalyzed reactions.

METHODS

This review synthesizes recent developments in AI-enhanced catalytic modelling, kinetic parameter estimation, and multi-scale reaction simulations and explores machine learning models such as Random Forest, Gradient Boosting, Artificial Neural Networks, and Gaussian Processes to predict key catalytic performance indicators. Additionally, high-throughput simulated screening and computational methods such as Density Functional Theory simulations and molecular descriptor-based modelling have been used to improve catalyst design tactics. Summary of the ML models which were trained and validated using open source frameworks such as scikit-learn, TensorFlow, and PyTorch is also presented in this paper. Most of the research studies datasets were using the resource data from Catalysis Hub and the materials project. Techniques for data processing and pre-processing include methods for choosing the component features, such as d-band center analysis, adsorption energy calculations, and algorithm normalization. This review study consists of an in-depth analysis of how data-driven modelling improves catalyst performance, and its prediction and optimization in hydrogenation catalysis reactions by artificial intelligence and machine learning driven approaches.

Cascade Catalytic Systems for Converting CO2 into C2+ Products

The excessive emission and continuous accumulation of CO2 have precipitated serious social and environmental issues. However, CO2 can also serve as an abundant, inexpensive, and non-toxic renewable C1 carbon source for synthetic reactions. To achieve carbon neutrality and recycling, it is crucial to convert CO2 into value-added products through chemical pathways. Multi-carbon (C2+) products, compared to C1 products, offer a broader range of applications and higher economic returns. Despite this, converting CO2 into C2+ products is difficult due to its stability and the high energy required for C-C coupling. Cascade catalytic reactions offer a solution by coordinating active components, promoting intermediate transfers, and facilitating further transformations. This method lowers energy consumption. Recent advancements in cascade catalytic systems have allowed for significant progress in synthesizing C2+ products from CO2. This review highlights the features and advantages of cascade catalysis strategies, explores the synergistic effects among active sites, and examines the mechanisms within these systems. It also outlines future prospects for CO2 cascade catalytic synthesis, offering a framework for efficient CO2 utilization and the development of next-generation catalytic systems.

Engineering highly active nuclease enzymes with machine learning and high-throughput screening

Optimizing enzymes to function in novel chemical environments is a central goal of synthetic biology, but optimization is often hindered by a rugged fitness landscape and costly experiments. In this work, we present TeleProt, a machine learning (ML) framework that blends evolutionary and experimental data to design diverse protein libraries, and employ it to improve the catalytic activity of a nuclease enzyme that degrades biofilms that accumulate on chronic wounds. After multiple rounds of high-throughput experiments, TeleProt found a significantly better top-performing enzyme than directed evolution (DE), had a better hit rate at finding diverse, high-activity variants, and was even able to design a high-performance initial library using no prior experimental data. We have released a dataset of 55,000 nuclease variants, one of the most extensive genotype-phenotype enzyme activity landscapes to date, to drive further progress in ML-guided design. A record of this paper's transparent peer review process is included in the supplemental information.

Biocatalytic oxyfunctionalization of unsaturated fatty acids to oxygenated chemicals via hydroxy fatty acids

The selective oxyfunctionalization of unsaturated fatty acids is difficult in chemical reactions, whereas regio- and stereoselective oxyfunctionalization is often performed in biocatalytic synthesis. Fatty acid oxygenases, including hydratases, lipoxygenases, dioxygenases, diol synthases, cytochrome P450 monooxygenases, peroxygenases, and 12-hydroxylases, are used to convert C16 and C18 unsaturated fatty acids to diverse regio- and stereoselective mono-, di-, and trihydroxy fatty acids via selective oxyfunctionalization. The formed hydroxy fatty acids or hydroperoxy fatty acids are metabolized to industrially important oxygenated chemicals such as lactones, green leaf volatiles, and bioplastic monomers, including ω-hydroxy fatty acids, α,ω-dicarboxylic acids, and fatty alcohols, by biocatalysts. For increased oxyfunctionalization of unsaturated fatty acids, enzyme engineering, functional and balanced expression in recombinant cells, selection of suitable catalyst types, and reaction engineering have been suggested. This review describes biocatalysts involved in the oxyfunctionalization of unsaturated fatty acids and the production of hydroxy fatty acids and oxygenated chemicals.

Unraveling the molecular basis of substrate specificity and halogen activation in vanadium-dependent haloperoxidases

Vanadium-dependent haloperoxidases (VHPOs) are biotechnologically valuable and operationally versatile biocatalysts. VHPOs share remarkable active-site structural similarities yet display variable reactivity and selectivity. The factors dictating substrate specificity and, thus, a general understanding of VHPO reaction control still need to be discovered. This work's strategic single-point mutation in the cyanobacterial bromoperoxidase AmVHPO facilitates a selectivity switch to allow aryl chlorination. This mutation induces loop formation that interacts with the neighboring protein monomer, creating a tunnel to the active sites. Structural analysis of the substrate-R425S-mutant complex reveals a substrate-binding site at the interface of two adjacent units. There, residues Glu139 and Phe401 interact with arenes, extending the substrate residence time close to the vanadate cofactor and stabilizing intermediates. Our findings validate the long-debated existence of direct substrate binding and provide a detailed VHPO mechanistic understanding. This work will pave the way for a broader application of VHPOs in diverse chemical processes.

Seven critical challenges in synthetic one-carbon assimilation and their potential solutions

Synthetic C1 assimilation holds the promise of facilitating carbon capture while mitigating greenhouse gas emissions, yet practical implementation in microbial hosts remains relatively limited. Despite substantial progress in pathway design and prototyping, most efforts stay at the proof-of-concept stage, with frequent failures observed even under in vitro conditions. This review identifies seven major barriers constraining the deployment of synthetic C1 metabolism in microorganisms and proposes targeted strategies for overcoming these issues. A primary limitation is the low catalytic activity of carbon-fixing enzymes, particularly carboxylases, which restricts the overall pathway performance. In parallel, challenges in expressing multiple heterologous genes-especially those encoding metal-dependent or oxygen-sensitive enzymes-further hinder pathway functionality. At the systems level, synthetic C1 pathways often exhibit poor flux distribution, limited integration with the host metabolism, accumulation of toxic intermediates, and disruptions in redox and energy balance. These factors collectively reduce biomass formation and compromise product yields in biotechnological setups. Overcoming these interconnected challenges is essential for moving synthetic C1 assimilation beyond conceptual stages and enabling its application in scalable, efficient bioprocesses towards a circular bioeconomy.

Engineering the Reaction Pathway of a Non-heme Iron Oxygenase Using Ancestral Sequence Reconstruction

Non-heme iron (FeII), α-ketoglutarate (α-KG)-dependent oxygenases are a family of enzymes that catalyze an array of transformations that cascade forward after the formation of radical intermediates. Achieving control over the reaction pathway is highly valuable and a necessary step toward broadening the applications of these biocatalysts. Numerous approaches have been used to engineer the reaction pathway of FeII/α-KG-dependent enzymes, including site-directed mutagenesis, DNA shuffling, and site-saturation mutagenesis, among others. Herein, we showcase a novel ancestral sequence reconstruction (ASR)-guided strategy in which evolutionary information is used to pinpoint the residues critical for controlling different reaction pathways. Following this, a combinatorial site-directed mutagenesis approach was used to quickly evaluate the importance of each residue. These results were validated using a DNA shuffling strategy and through quantum mechanical/molecular mechanical (QM/MM) simulations. Using this approach, we identified a set of active site residues together with a key hydrogen bond between the substrate and an active site residue, which are crucial for dictating the dominant reaction pathway. Ultimately, we successfully converted both extant and ancestral enzymes that perform benzylic hydroxylation into variants that can catalyze an oxidative ring-expansion reaction, showcasing the potential of utilizing ASR to accelerate the reaction pathway engineering within enzyme families that share common structural and mechanistic features.

Protein representations: Encoding biological information for machine learning in biocatalysis

Enzymes offer a more environmentally friendly and low-impact solution to conventional chemistry, but they often require additional engineering for their application in industrial settings, an endeavour that is challenging and laborious. To address this issue, the power of machine learning can be harnessed to produce predictive models that enable the in silico study and engineering of improved enzymatic properties. Such machine learning models, however, require the conversion of the complex biological information to a numerical input, also called protein representations. These inputs demand special attention to ensure the training of accurate and precise models, and, in this review, we therefore examine the critical step of encoding protein information to numeric representations for use in machine learning. We selected the most important approaches for encoding the three distinct biological protein representations - primary sequence, 3D structure, and dynamics - to explore their requirements for employment and inductive biases. Combined representations of proteins and substrates are also introduced as emergent tools in biocatalysis. We propose the division of fixed representations, a collection of rule-based encoding strategies, and learned representations extracted from the latent spaces of large neural networks. To select the most suitable protein representation, we propose two main factors to consider. The first one is the model setup, which is influenced by the size of the training dataset and the choice of architecture. The second factor is the model objectives such as consideration about the assayed property, the difference between wild-type models and mutant predictors, and requirements for explainability. This review is aimed at serving as a source of information and guidance for properly representing enzymes in future machine learning models for biocatalysis.

Accelerating enzyme discovery and engineering with high-throughput screening

Covering: up to August 2024Enzymes play an essential role in synthesizing value-added chemicals with high specificity and selectivity. Since enzymes utilize substrates derived from renewable resources, biocatalysis offers a pathway to an efficient bioeconomy with reduced environmental footprint. However, enzymes have evolved over millions of years to meet the needs of their host organisms, which often do not align with industrial requirements. As a result, enzymes frequently need to be tailored for specific industrial applications. Combining enzyme engineering with high-throughput screening has emerged as a key approach for developing novel biocatalysts, but several challenges are yet to be addressed. In this review, we explore emergent strategies and methods for isolating, creating, and characterizing enzymes optimized for bioproduction. We discuss fundamental approaches to discovering and generating enzyme variants and identifying those best suited for specific applications. Additionally, we cover techniques for creating libraries using automated systems and highlight innovative high-throughput screening methods that have been successfully employed to develop novel biocatalysts for natural product synthesis.

Machine learning-assisted amidase-catalytic enantioselectivity prediction and rational design of variants for improving enantioselectivity

Biocatalysis is an attractive approach for the synthesis of chiral pharmaceuticals and fine chemicals, but assessing and/or improving the enantioselectivity of biocatalyst towards target substrates is often time and resource intensive. Although machine learning has been used to reveal the underlying relationship between protein sequences and biocatalytic enantioselectivity, the establishment of substrate fitness space is usually disregarded by chemists and is still a challenge. Using 240 datasets collected in our previous works, we adopt chemistry and geometry descriptors and build random forest classification models for predicting the enantioselectivity of amidase towards new substrates. We further propose a heuristic strategy based on these models, by which the rational protein engineering can be efficiently performed to synthesize chiral compounds with higher ee values, and the optimized variant results in a 53-fold higher E-value comparing to the wild-type amidase. This data-driven methodology is expected to broaden the application of machine learning in biocatalysis research.

Revolutionizing Molecular Design for Innovative Therapeutic Applications through Artificial Intelligence

The field of computational protein engineering has been transformed by recent advancements in machine learning, artificial intelligence, and molecular modeling, enabling the design of proteins with unprecedented precision and functionality. Computational methods now play a crucial role in enhancing the stability, activity, and specificity of proteins for diverse applications in biotechnology and medicine. Techniques such as deep learning, reinforcement learning, and transfer learning have dramatically improved protein structure prediction, optimization of binding affinities, and enzyme design. These innovations have streamlined the process of protein engineering by allowing the rapid generation of targeted libraries, reducing experimental sampling, and enabling the rational design of proteins with tailored properties. Furthermore, the integration of computational approaches with high-throughput experimental techniques has facilitated the development of multifunctional proteins and novel therapeutics. However, challenges remain in bridging the gap between computational predictions and experimental validation and in addressing ethical concerns related to AI-driven protein design. This review provides a comprehensive overview of the current state and future directions of computational methods in protein engineering, emphasizing their transformative potential in creating next-generation biologics and advancing synthetic biology.

Discovery of alkaline laccases from basidiomycete fungi through machine learning-based approach

BACKGROUND

Laccases can oxidize a broad spectrum of substrates, offering promising applications in various sectors, such as bioremediation, biomass fractionation in future biorefineries, and synthesis of biochemicals and biopolymers. However, laccase discovery and optimization with a desirable pH optimum remains a challenge due to the labor-intensive and time-consuming nature of the traditional laboratory methods.

RESULTS

This study presents a machine learning (ML)-integrated approach for predicting pH optima of basidiomycete fungal laccases, utilizing a small, curated dataset against a vast metagenomic data. Comparative computational analyses unveiled the structural and pH-dependent solubility differences between acidic and neutral-alkaline laccases, helping us understand the molecular bases of enzyme pH optimum. The pH profiling of the two ML-predicted alkaline laccase candidates from the basidiomycete fungus Lepista nuda further validated our computational approach, showing the accuracy of this comprehensive method.

CONCLUSIONS

This study uncovers the efficacy of ML in the prediction of enzyme pH optimum from minimal datasets, marking a significant step towards harnessing computational tools for systematic screening of enzymes for biotechnology applications.

Annotation Vocabulary (Might Be) All You Need

Protein Language Models (pLMs) have revolutionized the computational modeling of protein systems, building numerical embeddings that are centered around structural features. To enhance the breadth of biochemically relevant properties available in protein embeddings, we engineered the Annotation Vocabulary, a transformer readable language of protein properties defined by structured ontologies. We trained Annotation Transformers (AT) from the ground up to recover masked protein property inputs without reference to amino acid sequences, building a new numerical feature space on protein descriptions alone. We leverage AT representations in various model architectures, for both protein representation and generation. To showcase the merit of Annotation Vocabulary integration, we performed 515 diverse downstream experiments. Using a novel loss function and only $3 in commercial compute, our premier representation model CAMP produces state-of-the-art embeddings for five out of 15 common datasets with competitive performance on the rest; highlighting the computational efficiency of latent space curation with Annotation Vocabulary. To standardize the comparison of de novo generated protein sequences, we suggest a new sequence alignment-based score that is more flexible and biologically relevant than traditional language modeling metrics. Our generative model, GSM, produces high alignment scores from annotation-only prompts with a BERT-like generation scheme. Of particular note, many GSM hallucinations return statistically significant BLAST hits, where enrichment analysis shows properties matching the annotation prompt - even when the ground truth has low sequence identity to the entire training set. Overall, the Annotation Vocabulary toolbox presents a promising pathway to replace traditional tokens with members of ontologies and knowledge graphs, enhancing transformer models in specific domains. The concise, accurate, and efficient descriptions of proteins by the Annotation Vocabulary offers a novel way to build numerical representations of proteins for protein annotation and design.

The Development and Opportunities of Predictive Biotechnology

Recent advances in bioeconomy allow a holistic view of existing and new process chains and enable novel production routines continuously advanced by academia and industry. All this progress benefits from a growing number of prediction tools that have found their way into the field. For example, automated genome annotations, tools for building model structures of proteins, and structural protein prediction methods such as AlphaFold2TM or RoseTTAFold have gained popularity in recent years. Recently, it has become apparent that more and more AI-based tools are being developed and used for biocatalysis and biotechnology. This is an excellent opportunity for academia and industry to accelerate advancements in the field further. Biotechnology, as a rapidly growing interdisciplinary field, stands to benefit greatly from these developments.

Biocatalysis of CO2 and CH4: Key enzymes and challenges

Mitigating greenhouse gas emissions is a critical challenge for promoting global sustainability. The utilization of CO2 and CH4 as substrates for the production of valuable products offers a promising avenue for establishing an eco-friendly economy. Biocatalysis, a sustainable process utilizing enzymes to facilitate biochemical reactions, plays a significant role in upcycling greenhouse gases. This review provides a comprehensive overview of the enzymes and associated reactions involved in the biocatalytic conversion of CO2 and CH4. Furthermore, the challenges facing the field are discussed, paving the way for future research directions focused on developing robust enzymes and systems for the efficient fixation of CO2 and CH4.

Enzymatic micro/nanomotors in biomedicine: from single motors to swarms

Micro/nanomotors (MNMs) have evolved from single self-propelled entities to versatile systems capable of performing one or multiple biomedical tasks. When single MNMs self-assemble into coordinated swarms, either under external control or triggered by chemical reactions, they offer advantages that individual MNMs cannot achieve. These benefits include intelligent multitasking and adaptability to changes in the surrounding environment. Here, we provide our perspective on the evolution of MNMs, beginning with the development of enzymatic MNMs since the first theoretical model was proposed in 2005. These enzymatic MNMs hold immense promise in biomedicine due to their advantages in biocompatibility and fuel availability. Subsequently, we introduce the design and application of single motors in biomedicine, followed by the control of MNM swarms and their biomedical applications. In the end, we propose viable solutions for advancing the development of MNM swarms and anticipate valuable insights into the creation of more intelligent and controllable MNM swarms for biomedical applications.

Effective engineering of a ketoreductase for the biocatalytic synthesis of an ipatasertib precursor

Semi-rational enzyme engineering is a powerful method to develop industrial biocatalysts. Profiting from advances in molecular biology and bioinformatics, semi-rational approaches can effectively accelerate enzyme engineering campaigns. Here, we present the optimization of a ketoreductase from Sporidiobolus salmonicolor for the chemo-enzymatic synthesis of ipatasertib, a potent protein kinase B inhibitor. Harnessing the power of mutational scanning and structure-guided rational design, we created a 10-amino acid substituted variant exhibiting a 64-fold higher apparent kcat and improved robustness under process conditions compared to the wild-type enzyme. In addition, the benefit of algorithm-aided enzyme engineering was studied to derive correlations in protein sequence-function data, and it was found that the applied Gaussian processes allowed us to reduce enzyme library size. The final scalable and high performing biocatalytic process yielded the alcohol intermediate with ≥ 98% conversion and a diastereomeric excess of 99.7% (R,R-trans) from 100 g L-1 ketone after 30 h. Modelling and kinetic studies shed light on the mechanistic factors governing the improved reaction outcome, with mutations T134V, A238K, M242W and Q245S exerting the most beneficial effect on reduction activity towards the target ketone.

Perspectives on Computational Enzyme Modeling: From Mechanisms to Design and Drug Development

Understanding enzyme mechanisms is essential for unraveling the complex molecular machinery of life. In this review, we survey the field of computational enzymology, highlighting key principles governing enzyme mechanisms and discussing ongoing challenges and promising advances. Over the years, computer simulations have become indispensable in the study of enzyme mechanisms, with the integration of experimental and computational exploration now established as a holistic approach to gain deep insights into enzymatic catalysis. Numerous studies have demonstrated the power of computer simulations in characterizing reaction pathways, transition states, substrate selectivity, product distribution, and dynamic conformational changes for various enzymes. Nevertheless, significant challenges remain in investigating the mechanisms of complex multistep reactions, large-scale conformational changes, and allosteric regulation. Beyond mechanistic studies, computational enzyme modeling has emerged as an essential tool for computer-aided enzyme design and the rational discovery of covalent drugs for targeted therapies. Overall, enzyme design/engineering and covalent drug development can greatly benefit from our understanding of the detailed mechanisms of enzymes, such as protein dynamics, entropy contributions, and allostery, as revealed by computational studies. Such a convergence of different research approaches is expected to continue, creating synergies in enzyme research. This review, by outlining the ever-expanding field of enzyme research, aims to provide guidance for future research directions and facilitate new developments in this important and evolving field.

Data-Driven Protein Engineering for Improving Catalytic Activity and Selectivity

Protein engineering is essential for altering the substrate scope, catalytic activity and selectivity of enzymes for applications in biocatalysis. However, traditional approaches, such as directed evolution and rational design, encounter the challenge in dealing with the experimental screening process of a large protein mutation space. Machine learning methods allow the approximation of protein fitness landscapes and the identification of catalytic patterns using limited experimental data, thus providing a new avenue to guide protein engineering campaigns. In this concept article, we review machine learning models that have been developed to assess enzyme-substrate-catalysis performance relationships aiming to improve enzymes through data-driven protein engineering. Furthermore, we prospect the future development of this field to provide additional strategies and tools for achieving desired activities and selectivities.