New research: Designing luciferase de novo using deep learning

Protein is the basis of life and the main executor of life functions. The sequence of amino acids determines its structure and function. Almost all proteins that can form a stable three-dimensional structure are natural proteins, and their amino acid sequences are formed by long-term natural evolution. When the structure and function of the natural protein cannot meet the needs of industrial or medical applications, it is necessary to design its structure and sequence to obtain a specific functional protein.

De novo protein design works reported internationally mainly use natural structural fragments as building blocks to splice and generate artificial structures. However, this method has shortcomings, such as single design results, and needs to be more sensitive to the details of the main chain structure, which limits the diversity and variability of the designed main chain structure. Therefore, using AI to "customize" new proteins has become the main solution to this problem.

A research team led by Professor David Baker of the University of Washington in the United States has developed an artificial intelligence algorithm based on deep learning - Family-wide Hallucination- used for protein structure prediction and design. The family-wide vision approach utilizes de novo sequence and structure-discovery capabilities to design protein loop structures and variable regions and structure-guided sequence optimization procedures to design protein core regions. David tried to use the deep learning algorithm Family-wide Hallucination to "tailor-made" a brand-new luciferase-LuxSit for luciferin DTZ and completed its de novo synthesis and performance optimization. Compared with natural luciferase, LuxSit exhibits superior activity, stability, and substrate specificity and can effectively catalyze substrate luminescence.

Structural data and protein sequence information of the existing luciferase family by deep learning, Family-wide Hallucination transforms this known "knowledge" to generate thousands of new and ideal protein three-dimensional structure skeletons with DTZ substrate-binding pocket-like structures and their corresponding amino acid sequences. The catalytic sites were then designed and screened with the help of Rosetta software, and finally, these candidate proteins were synthesized and expressed in E. coli. After cloning and activity identification, the luciferase with the strongest screening activity was named LuxSit by the researchers.

This research proves that algorithms based on deep learning can bring about earth-shaking changes in the field of protein engineering and may greatly promote the research and application of biological products or technologies such as enzyme preparations and biosensors.

Figure.1 Generation of idealized scaffolds and computational design of de novo luciferases.

Computer aided is the use of computer technology to understand and effectively utilize biological data, has become a cutting-edge applied science, and is crucial to today's biological research. Computational simulations have emerged as a powerful alternative and complement to experimental methods, enabling researchers to deepen their understanding of complex systems on computers.

Profacgen, recognizing the importance of computational technologies for pursuing advanced research in modern biology, established our own bioinformatics platform, which has emerged as a very sophisticated scientific infrastructure for bioinformatics involving state-of-the-art computational facilities, to offer support for both academic and industrial researchers to explore, understand and analyze complex biological data in a variety of biotechnological programs and projects.

• In Silico Protein-protein Interactions Prediction
• Bioinformatics Data Management
• Biological Data Analysis
• Computational Protein Analysis
• Custom Bioinformatics Software Development
• Virtual Screening
• Computer Aided Drug Design

This research proves that algorithms based on deep learning can bring about earth-shaking changes in the field of protein engineering and may greatly promote the research and application of biological products or technologies such as enzyme preparations and biosensors.

Profacgen has an experienced team that provides a wide range of services related to protein research. We have rich experience in protein research, advanced technical equipment, and professional researchers. If you want more service information, please feel free to contact us; Profacgen will be your trustworthy partner.

Reference

1. Yeh, A.HW., Norn, C., Kipnis, Y. et al. De novo design of luciferases using deep learning. Nature 614, 774–780 (2023). https://doi.org/10.1038/s41586-023-05696-3