Study Reveals Potential for New AI-Based Tools to Transform Enzyme Design, Catalyzing Cell-Free Bioproduction Across Multiple Industries
CoSaNN leverages advanced deep learning capabilities to generate new enzyme conformations based on the relationships between an enzyme's genetic sequence and its three-dimensional structure. (Credit: Hal Gatewood on Unsplash)
Enzymit, a specialty biochemical company developing cell-free enzymatic manufacturing technology, today announced the publication of a new study published on bioRxiv demonstrating the efficacy of its deep learning-based technology for novel enzyme design.
“The inherent instability of artificial enzymes, combined with production challenges and the limited range of reactions they can facilitate, has restricted their use in real-world applications, while the creation of novel enzymes has proved challenging due to the complex nature of such proteins,” said Gideon Lapidoth, PhD, CEO of Enzymit. “This research highlights the role of AI in overcoming the fundamental challenges developing commercially viable enzymes.”
The study, Context-Dependent Design of Induced-fit Enzymes using Deep Learning Generates Well Expressed, Thermally Stable and Active Enzymes, proposes an alternative approach to novel enzyme design, by modifying existing enzymes to work with new molecules under a variety of different conditions. This was achieved through the development of two new proprietary AI-based tools, CoSaNN (Conformation Sampling using Neural Network) and SolvIT.
CoSaNN leverages advanced deep learning capabilities to generate new enzyme conformations based on the relationships between an enzyme’s genetic sequence and its three-dimensional structure. This ability to redesign the shape of enzymes enabled new chemical reactions unattainable with current design tools. SolvIT, a graph neural network tool, served as a predictive model for protein solubility in the bacterium E. coli and provided an additional layer of optimization for producing highly expressed enzymes.
Combined, the tools enabled the creation of new enzymes with significantly higher thermal stability and exhibiting superior expression levels compared to alternative methods. 54% of the enzyme designs were successfully expressed in E. coli, of which 30% exhibited higher thermal stability than the template enzyme.
“This research demonstrates the transformative potential of artificial intelligence in novel enzyme sequence design, through its ability to capture complex interactions within a protein and to maintain its inherent mechanisms, even when heavily modified,” said Prof. Joseph Jacobson, board member and scientific advisor for Enzymit. “Harnessing the power of AI enables us to create vastly more stable enzymes in significantly greater volumes, and at significantly reduced costs for use in a wide range of commercial applications.”
Source: Company Press Release