Automation of a CRISPRi platform for enhanced isoprenol production in Pseudomonas putida [DBTL2]
Ontology highlight
ABSTRACT: Recent advances in genome engineering technologies have significantly enhanced the ability to perturb microbial metabolic networks. Despite this progress, many bioproduction efforts struggle to parse complex metabolic datasets (e.g., -omics) within the constraints of the traditional design-build-test-learn (DBTL) cycle to efficiently optimize titer, rate, and yield. Here, we consider bioproduction of isoprenol, a precursor to a sustainable aviation fuel blendstock, in Pseudomonas putida. Initially, a combination of heuristics and genome scale metabolic modeling was used to identify 120 genes associated with isoprenol production for CRISPR interference. We then developed a largely automated conversion pipeline to design multiplexed guide RNA arrays, construct strains harboring these arrays, and then assess proteomics-validated isoprenol production in P. putida, which reduced the DBT turnaround time from months using conventional methods to two weeks. By leveraging an active learning model, the Automated Recommendation Tool (ART), we systematically and rapidly identified new experimental designs that improved titer. ART successfully downselected from an initial design space of 2.1 x 108 possible combinations to approximately 300 combinations. After 6 successive DBTL cycles this ultimately yielded several different gRNA combinations with up to 5-fold titer improvement. Our conversion pipeline not only showcases the potential of integrating laboratory automation with machine learning, but also marks a significant advancement toward autonomous strain design.
INSTRUMENT(S):
ORGANISM(S): Pseudomonas Putida Kt2440
SUBMITTER:
Christopher Petzold
LAB HEAD: Christopher Petzold
PROVIDER: PXD063738 | Pride | 2025-09-19
REPOSITORIES: Pride
ACCESS DATA