Protein Engineering

Protein engineering, the process by which novel proteins with desired properties are developed, has grown by leaps and bounds since the first examples of protein mutagenesis were described over three decades ago. In the last 30 years, protein engineers have successfully created a wide range of proteins tailored to specific industrial, medical, and research applications. Yet challenges remain that prevent the engineering of complex protein functions on demand. With the aim of creating a forum for the protein engineering community, the inaugural Protein Engineering Canada Conference was held on June 20–22, 2014, in Ottawa, Canada. Over the course of the 2-day conference, 115 protein scientists from over 30 institutions representing 5 countries shared ideas, networked with colleagues, and learned about exciting new research covering a broad array of topics such as enzyme engineering, computational protein design, X-ray crystallography, protein NMR, molecular modeling, and protein evolution. This special issue highlights the breadth of research that was covered during the Protein Engineering Canada Conference by including previously unpublished research articles submitted by speakers as well as some selections that were submitted directly to Protein Science.

Bioprospecting lignin for biorefinery: Emerging innovations and strategies in microbial technology

The article discusses the development of recombinant protein engineering, primarily artificial proteins or de novo proteins, from the creation of the first proteins with a given spatial structure and biological activity to modern work in this area, which widely uses machine learning and artificial intelligence methods. The use of these methods, in particular the Rozetta and AlphaFold computer platforms, has led to tremendous progress in this area, as evidenced by last year’s Nobel Prize in Chemistry. Currently, these methods should be recommended for use in any modern laboratory conducting work on the physical chemistry of proteins and protein engineering. The article is based on the author’s report at a scientific session of the Division of Biological Sciences of the Russian Academy of Sciences on December 10, 2024.

Two popular methods exist to engineer a protein: directed evolution and rational design. Directed evolution utilizes a controlled environment to create proteins through induced mutations and selection, while rational design makes desired changes to a protein by directly manipulating its amino acids. Directed evolution is currently more commonly used, since rational design relies on structural knowledge of the protein of interest, which is often unavailable. Utilizing crowdsourcing manpower and computational power to improve protein depictions allows rational design to be more easily used to perform the manipulation of proteins. Two free programs, “Folding@home and “Foldit”, allow anyone with a computer and internet access to contribute to protein engineering. Folding@home relies on one’s computational power, while Foldit relies on user intuition to improve protein models. Rational design has allowed protein engineers to create artificial proteins that can be applied to the treatment of illnesses, research of enzyme activity in a living system, genetic engineering, and biological warfare. Starting with an overview of protein engineering, this paper discusses the methods of rational design and directed evolutions and goes on to explain how computer based programs can help in the advancement of rational design as a protein engineering method. Furthermore, this paper discusses the application of computer based programs in medicine and genetic engineering and presents some ethical issues that may arise from using such technology. The paper concludes with an analysis of whether or not computer based programs for protein engineering is worth the investment.

Protein sequence–function relationships are inherently complex, as amino acids at different positions can interact in highly unpredictable ways. A key question for protein evolution and engineering is how often epistasis extends beyond pairwise interactions to involve three or more positions. Although experimental data has accumulated rapidly in recent years, addressing this question remains challenging, as the number of possible interactions is typically enormous even for proteins of moderate size. Here, we introduce an interpretable machine learning framework for studying higher-order epistasis scalable to full-length proteins. Our model builds on the transformer architecture, with key modifications allowing us to assess the importance of higher-order interactions by fitting a series of models with increasing complexity. Applying our method to 10 large protein sequence-function datasets, we found that while additive effects explain the majority of the variance, within the epistatic component, the contribution of higher-order epistasis ranges from negligible to up to 60%. We also found higher-order epistasis is particularly important for generalizing locally sampled fitness data to distant regions of sequence space and for modeling an additional multi-peak fitness landscape. Our findings suggest that higher-order epistasis can play important roles in protein sequence-function relationships, and thus should be properly considered in protein engineering and evolutionary data analysis.

The introduction of heterologous pathways into microorganisms for high‐level synthesis of valuable compounds would be extremely meaningful. However, the endogenous heterologous pathway conversion point frequently serves as a bottleneck that limits product synthesis. (+)‐Nootkatone is an important functional sesquiterpenoid with a unique grapefruit flavor, which is frequently used as a fragrance. The low catalytic efficiency of the valencene synthase CnVS, the first heterologous enzyme in the biosynthesis of (+)‐nootkatone, is the major challenge that restricts the production of (+)‐nootkatone in Saccharomyces cerevisiae. In this study, we modified CnVS on the basis of the catalytic environment and mechanism of the enzyme. The residue M560 involving in alkyl transfer during substrate cyclization was found to affect activity. By tuning the residue, mutant M560L exhibited a 60% increase in the unit cell yield of (+)‐valencene, which is the precursor of (+)‐nootkatone. Furthermore, multistep strategies of metabolic engineering were employed that transferred the metabolic flux to the target product synthesis pathway and displayed the superimposed effect of metabolic and protein engineering on (+)‐nootkatone yield. Finally, multicopy integration of tHMG1 and ERG20ERG20‐GSG‐CnVSM560L boosted the production of (+)‐nootkatone to 85.43 mg/L in flask and 804.96 mg/L (5.29 mg/[L h]) in bioreactor. In conclusion, this study presents a paradigm for constructing microbial cell factories using a combination of protein and metabolic engineering in S. cerevisiae.

Protein engineering and design provide excellent tools to investigate the principles by which particular structural features relate to the mechanisms that underlie the biological function of a protein. In addition to studies aimed at dissecting the communication pathways within enzymes, recent advances in protein engineering approaches make it possible to generate enzymes with increased catalytic efficiency and specifically altered or newly introduced functions. Here, two approaches using state-of-the-art protein design and engineering are described in detail to demonstrate how key features of the myosin motor can be changed in a specific and predictable manner. First, it is shown how replacement of an actin-binding surface loop with synthetic sequences, whose flexibility and charge density is varied, can be employed to manipulate the actin affinity, the catalytic activity and the efficiency of coupling between actin- and nucleotide-binding sites of myosin motor constructs. Then the use of pre-existing molecular building blocks, which are derived from unrelated proteins, is described for manipulating the velocity and even the direction of movement of recombinant myosins.

The selective generation of covalent bonds between and within proteins would provide new avenues for studying protein function and engineering proteins with new properties. New covalent bonds were genetically introduced into proteins by enabling an unnatural amino acid (Uaa) to selectively react with a proximal natural residue. This proximity-enabled bioreactivity was expanded to a series of haloalkane Uaas. Orthogonal tRNA/synthetase pairs were evolved to incorporate these Uaas, which only form a covalent thioether bond with cysteine when positioned in close proximity. By using the Uaa and cysteine, spontaneous covalent bond formation was demonstrated between an affibody and its substrate Z protein, thereby leading to irreversible binding, and within the affibody to increase its thermostability. This strategy of proximity-enabled protein crosslinking (PEPC) may be generally expanded to target different natural amino acids, thus providing diversity and flexibility in covalent bond formation for protein research and protein engineering.

Protein sequence–function relationships are inherently complex, as amino acids at different positions can interact in highly unpredictable ways. A key question for protein evolution and engineering is how often epistasis extends beyond pairwise interactions to involve three or more positions. Although experimental data has accumulated rapidly in recent years, addressing this question remains challenging, as the number of possible interactions is typically enormous even for proteins of moderate size. Here, we introduce an interpretable machine learning framework for studying higher-order epistasis scalable to full-length proteins. Our model builds on the transformer architecture, with key modifications allowing us to assess the importance of higher-order interactions by fitting a series of models with increasing complexity. Applying our method to 10 large protein sequence-function datasets, we found that while additive effects explain the majority of the variance, within the epistatic component, the contribution of higher-order epistasis ranges from negligible to up to 60%. We also found higher-order epistasis is particularly important for generalizing locally sampled fitness data to distant regions of sequence space and for modeling an additional multi-peak fitness landscape. Our findings suggest that higher-order epistasis can play important roles in protein sequence-function relationships, and thus should be properly considered in protein engineering and evolutionary data analysis.

In this dispensation of the fourth industrial revolution, protein engineering has become a popular approach for increasing enzymatic activity, stability, and titer in the biosynthesis of natural products. This is attributed to its numerous advantages (over direct isolation from plants or via chemical synthesis), including decreasing or eliminating reaction byproducts, high precision, moderate handling of intricate and chemically unstable chemicals, overall reusability, and cost efficiency. Recently, protein engineering tools have advanced to redesign and enhance natural product biosynthesis. These methods include direct evolution, substrate engineering, medium engineering, enzyme engineering and immobilization, structure-assisted protein engineering, and advanced computational. Recent successes in implementing these emerging protein engineering technologies were critically discussed in this article. Also, the advantages, limitations, and applications in industrial and medical biotechnology were discussed. Last, future research directions and potential were also highlighted.

Modeling the impact of amino acid mutations on protein-protein interaction plays a crucial role in protein engineering and drug design. In this study, we develop GeoPPI, a novel structure-based deep-learning framework to predict the change of binding affinity upon mutations. Based on the three-dimensional structure of a protein, GeoPPI first learns a geometric representation that encodes topology features of the protein structure via a self-supervised learning scheme. These representations are then used as features for training gradient-boosting trees to predict the changes of protein-protein binding affinity upon mutations. We find that GeoPPI is able to learn meaningful features that characterize interactions between atoms in protein structures. In addition, through extensive experiments, we show that GeoPPI achieves new state-of-the-art performance in predicting the binding affinity changes upon both single- and multi-point mutations on six benchmark datasets. Moreover, we show that GeoPPI can accurately estimate the difference of binding affinities between a few recently identified SARS-CoV-2 antibodies and the receptor-binding domain (RBD) of the S protein. These results demonstrate the potential of GeoPPI as a powerful and useful computational tool in protein design and engineering. Our code and datasets are available at: https://github.com/Liuxg16/GeoPPI.

Key words islatravir - cascade reaction - biosynthesis - directed evolution - protein engineering - medicinal chemistry

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.

Machine learning-guided protein engineering is a new paradigm that enables the optimization of complex protein functions. Machine-learning methods use data to predict protein function without requiring a detailed model of the underlying physics or biological pathways. They accelerate protein engineering by learning from information contained in all measured variants and using it to select variants that are likely to be improved. We begin with a review of the basics of machine learning with a focus on applications to protein engineering and protein sequence-function datasets (Chapter 1). We used the entire machine-learning guided engineering paradigm to engineer the algal-derived light-gated channel channelrhodopsin (ChR), which can be used to modulate neuronal activity with light. We build models that discover ChRs with strong plasma membrane localization in mammalian cells (Chapter 2) and unprecedented light sensitivity and photocurrents for optogenetic applications (Chapter 3). Machine learning-guided evolution requires a machine-learning model that learns the relationship between sequence and function. For machine-learning models to learn about protein sequences, protein sequences must be represented as vectors or matrices of numbers. How each protein sequence is represented determines what can be learned. We learn continuous vector encodings of sequences from patterns in unlabeled sequences (Chapter 4). Learned encodings are low-dimensional, do not require alignments, and may improve performance by transferring information in unlabeled sequences to specific prediction tasks. Alternately, we demonstrate an interpretable Gaussian process kernel tailored to biological sequences (Chapter 6). In addition to a model to predict function from sequence, engineering requires a method to use the model to choose sequences for the next round of evolution. Most machine-learning guided engineering strategies assume that selected sequences can be queried directly. However, in directed evolution it is common to design a library of sequences and then sample stochastic batches from that library. We propose a batched stochastic Bayesian optimization algorithm for iteratively designing and screening site-saturation mutagenesis libraries (Chapter 5).

The imminent role of protein engineering in synthetic biology