Protein engineering advances unlock pathways to greener industrial biocatalysts

Their overview highlights innovative methods based on B-factor analysis, ancestral sequence reconstruction (ASR), and machine learning (ML), providing tools to design enzymes that withstand high temperatures and solvents. The review shows how these approaches can help develop robust biocatalysts for chemical, pharmaceutical, and food industries, enabling more sustainable and cost-effective manufacturing in the future.

Enzyme stability has long been a challenge for industrial applications, where reactions often require elevated temperatures, extended times, and organic solvents. Unlike chemical or metal catalysts, natural enzymes are prone to denaturation and reduced efficiency under such stresses. Protein engineers have therefore sought to replicate Darwinian evolution in the laboratory, beginning with random mutagenesis and advancing toward rational and semi-rational design. Over decades, approaches like error-prone PCR, DNA shuffling, and directed evolution have transformed enzyme engineering. Yet, bottlenecks in screening throughput and trade-offs between stability and catalytic activity persist. Against this backdrop, researchers have turned to computational and data-driven solutions, including structural analysis, evolutionary reconstructions, and ML algorithms, to guide mutagenesis and accelerate discovery.

A study (DOI:10.1016/j.bidere.2025.100005) published in BioDesign Research on 26 February 2025 by Bo Yuan’s, Hengquan Yang’s & Zhoutong Sun’s team, Shanxi University & Chinese Academy of Sciences, has created a powerful toolkit for improving enzyme thermostability and activity in industrial settings.

This review outlines three main avenues of progress. First, B-factor-based design uses atomic displacement data from crystallography or computational predictions to identify flexible protein regions prone to instability. By targeting these regions through mutagenesis, researchers have created enzymes with higher thermal tolerance, sometimes achieving over 400-fold increases in half-life. Tools like Rosetta and FoldX further refine predictions, while integration with ML enables identification of mutational “hotspots” at lower experimental costs. Second, ancestral sequence reconstruction (ASR) resurrects enzymes from extinct organisms, many of which exhibit naturally superior thermostability and broader substrate ranges. These ancestral enzymes provide stable templates for further optimization, as seen in alcohol dehydrogenases and laccases engineered for industrial use. Modern software such as FireProtASR, FastML, and PhyloBot makes ASR more accessible, lowering the technical barrier for experimentalists. Third, machine learning-driven protein design has emerged as a transformative tool. With vast protein databases like UniProt and PDB feeding deep learning models, algorithms can now predict enzyme properties, guide multi-site mutagenesis, and accelerate directed evolution. Studies using ML have delivered dramatic stability gains, including variants with 67-fold longer half-lives and significantly improved enantioselectivity. Advances in structural prediction tools like AlphaFold2 and RoseTTAFold further expand the scope of rational design. Together, these approaches illustrate a shift from trial-and-error mutagenesis toward predictive, computationally assisted engineering. The researchers emphasize that while each method has limitations—B-factors are resolution-dependent, ASR sequences carry uncertainties, and ML is constrained by dataset quality—their combined application is reshaping the field. Case studies highlighted in this review demonstrate how these methods not only improve thermostability but also help balance activity, selectivity, and cost-efficiency, traits essential for real-world applications. Importantly, the convergence of computational tools and experimental validation creates a feedback loop that accelerates discovery while reducing resource demands.

In summary, this review highlights how advances in protein engineering are driving the design of highly stable and efficient enzymes suitable for industrial environments. By integrating B-factor analysis, ancestral sequence reconstruction, and machine learning, researchers are overcoming long-standing barriers to enzyme stability. These strategies promise to transform industrial biocatalysis, offering pathways to greener manufacturing processes, reduced energy consumption, and novel applications across biotechnology.

###

References

DOI

10.1016/j.bidere.2025.100005

Original Source URL

https://doi.org/10.1016/j.bidere.2025.100005

Funding information

This work was supported by the National Key Research and Development Program of China (No. 2021YFC2101900 and No. 2021YFA0910400), the National Natural Science Foundation of China (No. 32171462 and No. 32301277), Tianjin Synthetic Biotechnology Innovation Capacity Improvement Project (No. TSBICIP-CXRC-040), the Natural Science Foundation of Tianjin (No. 21JCJQJC00110).

About BioDesign Research

BioDesign Research is dedicated to information exchange in the interdisciplinary field of biosystems design. Its unique mission is to pave the way towards the predictable de novo design and assessment of engineered or reengineered living organisms using rational or automated methods to address global challenges in health, agriculture, and the environment.