Investigating the phenotypic variation and proteomic response to lipid biosynthesis dysregulation across genetic backgrounds

Lipids are the core structural components of all cell membranes and organelles as well as crucial signaling molecules. In humans, dysregulation of lipid biosynthesis leads to numerous diseases including cancer, diabetes, heart disease and neurological diseases. Outside of diseases, altering the lipid levels of microorganisms is essential for industrial applications and metabolite production. Much of our knowledge regarding lipid metabolism has been discovered and characterised in the genetic model Saccharomyces cerevisiae (Baker’s yeast). Although, this knowledge mostly stems from a single genetic background, ignoring the inherent differences between individuals. My thesis therefore aims to understand the molecular variability underlying lipid biosynthesis perturbation across genetically diverse yeast strains using three lipid biosynthesis inhibitors: atorvastatin (sterol biosynthesis), cerulenin (fatty acid biosynthesis) and myriocin (sphingolipid biosynthesis).

In Chapter 2, I investigated the variation in growth inhibition in a library of 929 genetically diverse yeast strains in response to treatment with atorvastatin (sterol biosynthesis inhibitor), cerulenin (fatty acid synthesis inhibitor) and myriocin (sphingolipid biosynthesis inhibitor). The inhibition in growth was largely explained by grouping the strains into phylogenetic clades, whereby specific clades showed either increased resistance or sensitivity to each inhibitor. Of note were the alpechin and French Guiana human clades that were resistant to all treatments and contained a large number of introgressed genes from Saccharomyces paradoxus, the sister species to S. cerevisiae. Grouping the strains by ecological niches such as brewing, bioethanol production and human isolates, ploidy, or zygosity did not correlate strongly with growth inhibition. More specifically, copy number analysis identified potential genes, that when increased or decreased, were associated with either increased resistance or sensitivity to each treatment.

In Chapter 3, I investigated variation at the level of the proteome. Via selection for the ability to form discrete cells in liquid culture and having a haploid mating type of MATα, CBS1252 (bakery strain), CBS7765 (natural isolate) and S288C (lab strain) were mated with the yeast GFP-tagged protein collection, which consists of approximately 4,200 strains each with a different GFP-tagged protein. The resultant diploid strains as well as the haploid GFP-tagged strains were imaged using high-throughput confocal microscopy. Automated image analysis and machine learning were applied to identify changes in abundance and localisation of proteins. Prompted by increases in lipid biosynthesis proteins, thin-layer chromatography and lipid droplet staining showed an increase in neutral lipids in CBS1252 as well as differences in lipid droplet counts per cell in each background. Based on increased mitochondrial protein abundance in CBS1252, altered mitochondrial protein distribution in the haploid S288C, and increased mitochondrial membrane potential in CBS1252, it is plausible that these background-specific mitochondrial changes correlate with background-specific changes in lipid metabolism.

In Chapter 4, I treated the diploid strains constructed in Chapter 3 with atorvastatin, cerulenin, and myriocin and imaged the strains using high throughput confocal microscopy. Via automated image analysis and machine learning to identify changes in protein abundance and localisation, most of the proteins that were upregulated in response to each treatment were unique to a single background. A greater degree of overlap was seen between backgrounds when investigating changes in protein localisation, with most changes observed in more than one background. The machine learning identified decreased lipid droplet numbers in atorvastatin-treated cells that coincided with decreased levels of sterol ester, lanosterol and triacylglycerol as well as an increase in free fatty acids. In conjunction with altered lipid levels, eisosome distribution, which relies on sterols and sphingolipid levels, were affected across all treatments and backgrounds. Also identified was a shift of the lipid droplet protein Pdr16 to the nucleus when treated with atorvastatin.

Overall, these results provide further insight into the variability of responses to lipid biosynthesis inhibition at the protein level across genetically diverse yeast strains. Notably, I identified strains that were both sensitive and resistant to lipid biosynthesis inhibitors and related these to genetic variation. I determined changes in protein abundance in response to lipid biosynthesis inhibition that were unique to specific genetic backgrounds as well as changes that were common to all backgrounds in response to lipid biosynthesis dysregulation. My results provide further insight into how cells respond to lipid biosynthesis inhibition, which is of interest in terms of understanding fundamental eukaryotic lipid metabolism, individual drug response in humans, and yeast-based genetic engineering of lipid metabolism in industrial applications.