Yeasts and fungi can be powerful platforms for production of antibiotics and other small molecules. Yet, rewiring yeasts a priori for optimal production is not possible, necessitating an iterative design-build-test-learn cycle. Within this framework, large-scale combinatorial libraries can be used to rapidly learn design rules to guide subsequent iterations. To enable this process, standards and methods for achieving rapid assembly of varied levels of expression, while preserving genetic stability, must be developed.
To this end, the first part of the current study advances S. cerevisiae parts library design, characterization, and analysis. Employing high-throughput techniques and automated liquid handling, the expression of over a thousand natural and synthetic yeast transcription units under different growth conditions can be analyzed via flow cytometry. With this approach, the combined effects of promoters and terminators on gene expression can be modeled. Our initial analyses point to a well-defined expression space that can be modeled with a simple second-order expression.
The second part of the current study demonstrates how parts characterization on this scale enables construction of combinatorial libraries of multi-gene pathways without repeating parts within a design. Coupled with standardized assembly techniques, a library of nearly two hundred variants of a six-gene pathway were constructed in a matter of weeks. The pathways were subsequently assayed for performance and, using statistical analysis, bottlenecks and optimal relative expression levels for the pathway were determined.
With these advances, we demonstrate a large-scale design-build-test-learn cycle that can be generally applied to any multi-gene system in yeast.