Predicting protein stability and solubility changes upon mutations: data perspective

Transfer learning to leverage larger datasets for improved prediction of protein stability changes

Amino acid mutations that lower a protein's thermodynamic stability are implicated in numerous diseases, and engineered proteins with enhanced stability can be important in research and medicine. Computational methods for predicting how mutations perturb protein stability are, therefore, of great interest. Despite recent advancements in protein design using deep learning, in silico prediction of stability changes has remained challenging, in part due to a lack of large, high-quality training datasets for model development. Here, we describe ThermoMPNN, a deep neural network trained to predict stability changes for protein point mutations given an initial structure. In doing so, we demonstrate the utility of a recently released megascale stability dataset for training a robust stability model. We also employ transfer learning to leverage a second, larger dataset by using learned features extracted from ProteinMPNN, a deep neural network trained to predict a protein's amino acid sequence given its three-dimensional structure. We show that our method achieves state-of-the-art performance on established benchmark datasets using a lightweight model architecture that allows for rapid, scalable predictions. Finally, we make ThermoMPNN readily available as a tool for stability prediction and design.

Machine Learning-Guided Protein Engineering

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

Advancing Enzyme's Stability and Catalytic Efficiency through Synergy of Force-Field Calculations, Evolutionary Analysis, and Machine Learning

Thermostability is an essential requirement for the use of enzymes in the bioindustry. Here, we compare different protein stabilization strategies using a challenging target, a stable haloalkane dehalogenase DhaA115. We observe better performance of automated stabilization platforms FireProt and PROSS in designing multiple-point mutations over the introduction of disulfide bonds and strengthening the intra- and the inter-domain contacts by in silico saturation mutagenesis. We reveal that the performance of automated stabilization platforms was still compromised due to the introduction of some destabilizing mutations. Notably, we show that their prediction accuracy can be improved by applying manual curation or machine learning for the removal of potentially destabilizing mutations, yielding highly stable haloalkane dehalogenases with enhanced catalytic properties. A comparison of crystallographic structures revealed that current stabilization rounds were not accompanied by large backbone re-arrangements previously observed during the engineering stability of DhaA115. Stabilization was achieved by improving local contacts including protein-water interactions. Our study provides guidance for further improvement of automated structure-based computational tools for protein stabilization.

Transfer learning to leverage larger datasets for improved prediction of protein stability changes

Amino acid mutations that lower a protein's thermodynamic stability are implicated in numerous diseases, and engineered proteins with enhanced stability are important in research and medicine. Computational methods for predicting how mutations perturb protein stability are therefore of great interest. Despite recent advancements in protein design using deep learning, in silico prediction of stability changes has remained challenging, in part due to a lack of large, high-quality training datasets for model development. Here we introduce ThermoMPNN, a deep neural network trained to predict stability changes for protein point mutations given an initial structure. In doing so, we demonstrate the utility of a newly released mega-scale stability dataset for training a robust stability model. We also employ transfer learning to leverage a second, larger dataset by using learned features extracted from a deep neural network trained to predict a protein's amino acid sequence given its three-dimensional structure. We show that our method achieves competitive performance on established benchmark datasets using a lightweight model architecture that allows for rapid, scalable predictions. Finally, we make ThermoMPNN readily available as a tool for stability prediction and design.

Recent Advances in β-Glucosidase Sequence and Structure Engineering: A Brief Review

β-glucosidases (BGLs) play a crucial role in the degradation of lignocellulosic biomass as well as in industrial applications such as pharmaceuticals, foods, and flavors. However, the application of BGLs has been largely hindered by issues such as low enzyme activity, product inhibition, low stability, etc. Many approaches have been developed to engineer BGLs to improve these enzymatic characteristics to facilitate industrial production. In this article, we review the recent advances in BGL engineering in the field, including the efforts from our laboratory. We summarize and discuss the BGL engineering studies according to the targeted functions as well as the specific strategies used for BGL engineering.

In-depth analysis of biocatalysts by microfluidics: An emerging source of data for machine learning

Nowadays, the vastly increasing demand for novel biotechnological products is supported by the continuous development of biocatalytic applications which provide sustainable green alternatives to chemical processes. The success of a biocatalytic application is critically dependent on how quickly we can identify and characterize enzyme variants fitting the conditions of industrial processes. While miniaturization and parallelization have dramatically increased the throughput of next-generation sequencing systems, the subsequent characterization of the obtained candidates is still a limiting process in identifying the desired biocatalysts. Only a few commercial microfluidic systems for enzyme analysis are currently available, and the transformation of numerous published prototypes into commercial platforms is still to be streamlined. This review presents the state-of-the-art, recent trends, and perspectives in applying microfluidic tools in the functional and structural analysis of biocatalysts. We discuss the advantages and disadvantages of available technologies, their reproducibility and robustness, and readiness for routine laboratory use. We also highlight the unexplored potential of microfluidics to leverage the power of machine learning for biocatalyst development.

Target-template relationships in protein structure prediction and their effect on the accuracy of thermostability calculations

Improving protein thermostability has been a labor- and time-consuming process in industrial applications of protein engineering. Advances in computational approaches have facilitated the development of more efficient strategies to allow the prioritization of stabilizing mutants. Among these is FEP+, a free energy perturbation implementation that uses a thoroughly tested physics-based method to achieve unparalleled accuracy in predicting changes in protein thermostability. To gauge the applicability of FEP+ to situations where crystal structures are unavailable, here we have applied the FEP+ approach to homology models of 12 different proteins covering 316 mutations. By comparing predictions obtained with homology models to those obtained using crystal structures, we have identified that local rather than global sequence conservation between target and template sequence is a determining factor in the accuracy of predictions. By excluding mutation sites with low local sequence identity (<40%) to a template structure, we have obtained predictions with comparable performance to crystal structures (R2 of 0.67 and 0.63 and an RMSE of 1.20 and 1.16 kcal/mol for crystal structure and homology model predictions, respectively) for identifying stabilizing mutations when incorporating residue scanning into a cascade screening strategy. Additionally, we identify and discuss inherent limitations in sequence alignments and homology modeling protocols that translate into the poor FEP+ performance of a few select examples. Overall, our retrospective study provides detailed guidelines for the application of the FEP+ approach using homology models for protein thermostability predictions, which will greatly extend this approach to studies that were previously limited by structure availability.

Exploring the Effect of Enhanced Sampling on Protein Stability Prediction

Changes in protein stability due to side-chain mutations are evaluated by alchemical free-energy calculations based on classical molecular dynamics (MD) simulations in explicit solvent using the GROMOS force field. Three proteins of different complexity with a total number of 93 single-point mutations are analyzed, and the relative free-energy differences are discussed with respect to configurational sampling and (dis)agreement with experimental data. For the smallest protein studied, a 34-residue WW domain, the starting structure dependence of the alchemical free-energy changes, is discussed in detail. Deviations from previous simulations for the two other proteins are shown to result from insufficient sampling in the earlier studies. Hamiltonian replica exchange in combination with multiple starting structures and sufficient sampling time of more than 100 ns per intermediate alchemical state is required in some cases to reach convergence.

Toward Compilation of Balanced Protein Stability Data Sets: Flattening the ΔΔG Curve through Systematic Enrichment

Often studies analyzing stability data sets and/or predictors ignore neutral mutations and use a binary classification scheme labeling only destabilizing and stabilizing mutations. Recognizing that highly concentrated neutral mutations interfere with data set quality, we have explored three protein stability data sets: S2648, PON-tstab, and the symmetric S^sym that differ in size and quality. A characteristic leptokurtic shape in the ΔΔG distributions of all three data sets including the curated and symmetric ones was reported due to concentrated neutral mutations. To further investigate the impact of neutral mutations on ΔΔG predictions, we have comprehensively assessed the performance of 11 predictors on the PON-tstab data set. Correlation and error analyses showed that all of the predictors performed the best on the neutral mutations, while their performance became gradually worse as the ΔΔG of the mutations departed further from the neutral zone regardless of the direction, implying a bias toward dense mutations. To this end, after unraveling the role of concentrated neutral mutations in biases of stability data sets, we described a systematic enrichment approach to balance the ΔΔG distributions. Before enrichment, mutations were clustered based on their biochemical and/or structural features, and then three mutations were selected from every 2 kcal/mol of each cluster. Upon implementation of this approach by distinct clustering schemes, we generated five subsets varying in size and ΔΔG distributions. All subsets showed improved ΔΔG and frequency distributions. We ultimately reported that the errors toward enriched subsets were higher than those toward the parent data sets, confirming the enrichment of difficult-to-predict mutations in the subsets. In summary, we elaborated the prediction bias toward a concentrated neutral zone and also implemented a rational strategy to tackle this and other forms of biases. Ultimately, this study equipping us with an extended view of shortcomings of stability data sets is a step taken toward development of an unbiased predictor.

Data set and fitting dependencies when estimating protein mutant stability: Toward simple, balanced, and interpretable models

Accurate prediction of protein stability changes upon mutation (ΔΔG) is increasingly important to evolution studies, protein engineering, and screening of disease-causing gene variants but is challenged by biases in training data. We investigated 45 linear regression models trained on data sets that account systematically for destabilization bias and mutation-type bias B_M . The models were externally validated on three test data sets probing different pathologies and for internal consistency (symmetry and neutrality). Model structure and performance substantially depended on training data and even fitting method. We developed two final models: SimBa-IB for typical natural mutations and SimBa-SYM for situations where stabilizing and destabilizing mutations occur to a similar extent. SimBa-SYM, despite is simplicity, is essentially non-biased (vs. the S^sym data set) while still performing well for all data sets (R ~ 0.46-0.54, MAE = 1.16-1.24 kcal/mol). The simple models provide advantage in terms of interpretability, use and future improvement, and are freely available on GitHub.

Site-wise Diversification of Combinatorial Libraries Using Insights from Structure-guided Stability Calculations

Many auspicious clinical and industrial accomplishments have improved the human condition by means of protein engineering. Despite these achievements, our incomplete understanding of the sequence-structure-function relationship prevents rapid innovation. To tackle this problem, we must develop and integrate new and existing technologies. To date, directed evolution and rational design have dominated as protein engineering principles. Even so, prior to screening for novel or improved functions, a large collection of variants, within a protein library, exist along an ambiguous mutational terrain. Complicating things further, the choice of where to initialize investigation along a vast sequence space becomes even more difficult given that the majority of any sequence lacks function entirely. Unfortunately, even when considering functionally relevant positions, random substitutions can prove to be destabilizing, causing a hindrance to an otherwise function-inducing, stability-reliant folding process. To enhance productivity in the field, we seek to address this issue of destabilization, and subsequent disfunction, at protein-protein and protein-ligand interacting regions. Herein, the process of choosing amenable positions - and amino acids at those positions - allows for a refined, knowledge-based approach to combinatorial library design. Using structural data, we perform computational stability prediction with FoldX's PositionScan and Rosetta's ddG_monomer in tandem, allowing for the refinement of our thermodynamic stability data through the comparison of results. In turn, we provide a process for selecting in silico predicted mutually stabilizing positions and avoiding overly destabilizing ones that guides the site-wise diversification of combinatorial libraries.

SCONES: Self-Consistent Neural Network for Protein Stability Prediction Upon Mutation

Engineering proteins to have desired properties by mutating amino acids at specific sites is commonplace. Such engineered proteins must be stable to function. Experimental methods used to determine stability at throughputs required to scan the protein sequence space thoroughly are laborious. To this end, many machine learning based methods have been developed to predict thermodynamic stability changes upon mutation. These methods have been evaluated for symmetric consistency by testing with hypothetical reverse mutations. In this work, we propose transitive data augmentation, evaluating transitive consistency with our new S^transitive data set, and a new machine learning based method, the first of its kind, that incorporates both symmetric and transitive properties into the architecture. Our method, called SCONES, is an interpretable neural network that predicts small relative protein stability changes for missense mutations that do not significantly alter the structure. It estimates a residue's contributions toward protein stability (ΔG) in its local structural environment, and the difference between independently predicted contributions of the reference and mutant residues is reported as ΔΔG. We show that this self-consistent machine learning architecture is immune to many common biases in data sets, relies less on data than existing methods, is robust to overfitting, and can explain a substantial portion of the variance in experimental data.

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.

TISIGNER.com: web services for improving recombinant protein production

Experiments that are planned using accurate prediction algorithms will mitigate failures in recombinant protein production. We have developed TISIGNER (https://tisigner.com) with the aim of addressing technical challenges to recombinant protein production. We offer three web services, TIsigner (Translation Initiation coding region designer), SoDoPE (Soluble Domain for Protein Expression) and Razor, which are specialised in synonymous optimisation of recombinant protein expression, solubility and signal peptide analysis, respectively. Importantly, TIsigner, SoDoPE and Razor are linked, which allows users to switch between the tools when optimising genes of interest.

Web-based tools for computational enzyme design

Enzymes are in high demand for very diverse biotechnological applications. However, natural biocatalysts often need to be engineered for fine-tuning their properties towards the end applications, such as the activity, selectivity, stability to temperature or co-solvents, and solubility. Computational methods are increasingly used in this task, providing predictions that narrow down the space of possible mutations significantly and can enormously reduce the experimental burden. Many computational tools are available as web-based platforms, making them accessible to non-expert users. These platforms are typically user-friendly, contain walk-throughs, and do not require deep expertise and installations. Here we describe some of the most recent outstanding web-tools for enzyme engineering and formulate future perspectives in this field.

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.