Empowering yeast protein cell factories: biosensor & organelle biomimicry

Besides its long-history application in providing humanity with food and beverages, nowadays the baker’s yeast (Saccharomyces cerevisiae) has acquired the updated role as a popular microbial host in the production of valuable recombinant proteins for pharmaceutical and industrial usage. Its advantages as a protein factory include its generally regarded as safe (GRAS) status, well-understood genome, eukaryotic protein folding and modification, robustness in industrial-scale processes, resistance to contamination, protein secretion pathway, and rich molecular toolkits for genetic engineering. Constant efforts have been made to increase the production titer of yeast protein cell factories, including expression engineering at the recombinant DNA level to build the optimal protein expression cassettes, as well as strain engineering to improve the protein production capacity of the cell. The optimal engineering strategy is often protein-specific, therefore frequently requires large scale screening of genetic variables such as promoters, signal peptides and gene copy numbers, and screening of various yeast strains to determine the best combination for a certain target protein of interest (POI).

The conventional methodology to evaluate POI yield from different engineering strategies are mainly based on antibody or enzymatic activity of POI, which are slow, expensive, low through-put and protein-specific. On the other hand, the overproduction of a POI often triggers the unfolded protein response (UPR), the intensity of which has inherent correlation to the protein secretion titre. Taking advantage of this, we developed synthetic biosensors for the detection and measurement of the UPR in S. cerevisiae, using synthetic minimal genetic parts. Compared with conventional sensor designs using native and semi-synthetic UPR promoters, the synthetic minimal UPR sensors are not only concise in design, but also outperformed the previously reported sensors in resolution, dynamic range and signal intensity. We then demonstrated that our UPR sensors were able to distinguish between expression and strain engineering strategies that led to different POI titres, in a fast, cheap, high through-put and POI independent manner. Therefore, they can potentially be applied as a universal screening tool at the early stage of industrial development, to help determine the optimal engineering strategy to maximize the final POI yield.

To develop S. cerevisiae strains with better POI secretion capacity, the main-stream strategies are focused on the modification of machinery molecules directly or indirectly involved in the processing of POI, such as folding chaperones, molecules involved in disulfide bond isomerization, glycosylation, protein quality control and transport. While the subcellular infrastructures, including the morphology and organization of organelles, have been rarely included as strain engineering targets; despite their fundamental roles in providing the environment and scaffold that support the machinery molecules and the concomitant impact on POI production. The significance of infrastructure is best demonstrated by comparison in the morphology of the endoplasmic reticulum (ER) sheet, which is simple and a single cisterna in yeast, but is highly stacked and rough ER in mammalian professional protein secretory cells, such as the plasma cell.

Inspired by the ER of the plasma cell, we aimed to engineer the ER morphology of S. cerevisiae to mimic the plasma cell, by using various known ER shaping proteins from higher eukaryotes. Synthetic biology methodologies were applied to extract and augment the functional minimal modules from the original ER shaping proteins, and their effect on yeast ER morphology were systematically studied. We found that “ER proliferation coupled smooth ER stacks” morphology can be steadily induced by synthetic constructs containing a minimal non-specific transmembrane domain (TMD) plus a non-specific oligomerizing cytosolic domain. Based on the human ribosome binding protein Rrbp1, we built a synthetic construct creating stacked rough ER sheet in yeast. This synthetic version of Rrbp1 was significantly smaller than the original protein, and had improved ER shaping efficacy via the incorporation of a yeast native TMD and synthetic biophysical parts such as the superfolder GFP. The expression of another human ER shaping protein Ckap4 triggered stacked smooth lumen expanded ER with low stability in S. cerevisiae, where the N-terminus cytosolic domain, especially the first 32 amino acid header sequence had a profound impact on the Ckap4 type ER morphology. Using a modified yeast native TMD could improve the stability of this ER morphology. We found different categories of ER shaping proteins were engaged in a “turf war” in determining the ER morphology when co-expressed, with one category often dominant over the other. This study provides insights into the evolution and differentiation of ER morphology, as well as presenting another opportunity to engineer S. cerevisiae strains with improved protein secretion. It heralds a novel area of yeast engineering via systematic alteration of organelle morphology towards a predefined target, thus we would like to coin the word “organelle biomimicry”. 

This thesis has developed both a synthetic minimal UPR sensor as a novel tool for production evaluation and organelle biomimicry as a novel direction for strain improvement, both of which should strongly contribute to the empowerment of S. cerevisiae as a protein cell factory.