Recently, a research article titled "Design of Strictly Orthogonal Biosensors for Maximizing Renewable Biofuel Overproduction" by the team of Professor Huo Yixin from the School of Life Sciences at Beijing Institute of Technology was published in the top-tier journal "Journal of Advanced Research" (Zone 1). This study addresses the relationship between signal molecule promiscuity and industrial orthogonality by developing a machine learning-based design method. Taking the transcription factor BmoR as an example, the method successfully narrowed the engineering region of BmoR, significantly accelerating the acquisition of ideal mutants. It provides a new paradigm for the rational design of highly specific tools in synthetic biology and biofabrication. The work was conducted with Beijing Institute of Technology as the first corresponding institution, with postgraduate student Wu Tong (2020 cohort) as the first author, and Associate Professor Chen Zhenya and Professor Huo Yixin as co-corresponding authors.

Currently, transcription factor (TF) engineering relies on the combination of random mutagenesis and high-throughput screening. However, mutations in large regions can partially impair the function of transcription factors, while the presence of nonfunctional mutants reduces the screening efficiency for effective variants. By focusing the modification efforts on the functional domains that control signal molecule orthogonality, sensitivity, and detection range, the rate of obtaining ideal mutants can be significantly improved. Therefore, identifying these key regions is crucial for the efficient design of high-performance transcription factors. Artificial intelligence has driven advancements in synthetic biology and metabolic engineering, particularly through the application of machine learning in protein and enzyme engineering. Combining high-throughput screening with machine learning can accelerate the evaluation of large numbers of mutant performances, thereby substantially reducing experimental workload and costs. Based on this, this study establishes a semi-rational design strategy using machine learning by targeting the key regions affecting the binding of BmoR to signal molecules, enabling BmoR to achieve strict specificity in response to specific signal molecules.

Taking BmoR as an example, this study developed a method to quantify the influence of transcriptional activators, guiding the design of mutants with precise signal molecule recognition capabilities. A semi-rational design strategy assisted by machine learning was established using the random forest algorithm. By constructing a predictive model with 88.5% accuracy, three key regions significantly affecting the specific binding of BmoR to signal molecules were precisely identified, totaling 36 amino acid residues

By conducting semi-rational design on these three key regions, BmoR successfully achieved strict discrimination of isomers and structurally similar analogs. This study further demonstrated the stringent specificity of BmoR mutants toward n-amyl alcohol or isoamyl alcohol by measuring the affinity between BmoR and signaling molecules. Through simulation analysis of structural changes in the hexamer before and after BmoR mutation, the key residues influencing BmoR's recognition of signaling molecules were identified. Subsequently, utilizing a biosensing system with strict specificity for isoamyl alcohol, this work successfully screened a high-yield isoamyl alcohol-producing strain, achieving a biosynthetic yield of 12.6 g/L, the highest reported to date. The method combines machine learning with molecular simulation to optimize transcription factor engineering, addressing the issue of transcription factor promiscuity in signaling molecules and improving the detection accuracy of biosensors.

This work was supported by the National Natural Science Foundation of China, the Hebei Provincial Natural Science Foundation, the Tangshan Science and Technology Project, as well as the Public Experimental Center of Bioengineering and Medicine at Beijing Institute of Technology.