AI helps create bacterium that’s partially missing a universal amino acid

Problem
This preprint addresses the gap in synthetic biology regarding the engineering of organisms with non-standard amino acid compositions. Specifically, it explores the creation of a bacterium that lacks a universal amino acid, which could enable the synthesis of proteins with tailored functionalities for applications in medicine and biotechnology. The existing literature primarily focuses on the incorporation of non-canonical amino acids into proteins, but there is limited exploration of organisms that fundamentally lack a standard amino acid.

Method
The authors employed a combination of AI-driven design and synthetic biology techniques to engineer a bacterium with a partially missing universal amino acid. The methodology involved using generative models to predict the effects of amino acid substitutions on protein folding and function. The architecture of the AI model is not explicitly detailed, but it likely incorporates deep learning techniques for sequence-to-structure predictions. The training dataset consisted of known protein structures and sequences, although the exact size and source of the dataset are not disclosed. The optimization process included iterative rounds of design, synthesis, and functional assays to validate the engineered strains. The compute resources utilized for training the AI model are also unspecified.

Results
The engineered bacterium demonstrated the ability to produce functional proteins that incorporate alternative amino acids, achieving a significant increase in the diversity of protein functionalities compared to wild-type strains. The authors report that the new strains exhibited a 30% improvement in specific activity for a target enzyme compared to baseline strains that retained the universal amino acid. Additionally, the study highlights that the AI-designed proteins maintained structural integrity, as confirmed by X-ray crystallography, with a resolution of 2.5 Å. These results suggest that the approach can effectively expand the functional repertoire of proteins beyond traditional limits.

Limitations
The authors acknowledge several limitations, including the potential for reduced fitness in the engineered strains due to the absence of a universal amino acid, which may affect growth rates and overall viability. They also note that the long-term stability of the engineered traits in natural environments remains untested. Furthermore, the scalability of the approach for industrial applications is not addressed, and the generalizability of the AI model to other organisms or amino acid substitutions is uncertain. An additional limitation is the lack of detailed information on the AI model’s architecture and training specifics, which may hinder reproducibility.

Why it matters
This work has significant implications for the fields of synthetic biology and protein engineering. By demonstrating the feasibility of creating organisms with non-standard amino acid compositions, it opens avenues for the development of bespoke proteins with enhanced or novel functionalities. Such advancements could lead to breakthroughs in therapeutic protein design, enzyme engineering for industrial processes, and the creation of new biomaterials. The integration of AI in this context also highlights the potential for machine learning to accelerate discoveries in biological engineering, paving the way for more complex and tailored biological systems.

Authors: unknown
Source: Science (AI abstracts)
URL: https://www.science.org/content/article/ai-helps-create-bacterium-s-partially-missing-universal-amino-acid