However, the genetic basis for these advantageous phenotypes remains poorly understood. Therefore, we performed transcriptomics of D. hansenii under nitrogen starvation, iron starvation, and salt stress and compared the results to the same data from the model oleaginous yeast Yarrowia lipolytica and the related yeast Debaryomyces subglobosus. To compare the data, we first constructed a cross-species metabolic network and mapped the differential gene expression from each yeast in each condition. Second, we performed a coexpression analysis on the data using unsupervised machine learning. Both methods yielded valuable insights. The metabolic network analysis reaffirmed that Yarrowia upregulates lipid biosynthesis pathways under nitrogen-limiting conditions, and revealed that Debaryomyces yeasts have the highest expression of lipid metabolism in the unstressed condition. The network also showed that riboflavin overproduction in Debaryomyces species during salt stress likely results from overflow metabolism redirected toward flavin cofactor synthesis. Furthermore, the coexpression analysis revealed clusters of genes that were activated by stress. In these clusters, we identified stress-responsive gene modules, including known and putative transcription factors and membrane transporters. This integrative analysis enhances our understanding of the genotype-to-phenotype relationship in extremophilic yeasts and illustrates the power of combining network and data-driven methods to extract biological insight from cross-species transcriptomic data. It also provides metabolic and genetic evidence that argues that Debaryomyces hansenii is a promising host for metabolic engineering.
To demonstrate this, we then performed combinatorial metabolic engineering in D. hansenii for production of alkanes from the major constituents of lignocellulosic biomass - glucose, xylose, and arabinose. First, we used the genomic data from the comparative transcriptomics to derive functional D. hansenii promoters and terminators and clone them into a modular DNA assembly pipeline. We then built 18 alkane biosynthesis pathways with two thioesterase (TES) variants, three carboxylic acid reductase (CAR) variants, and three aldehyde deformylating oxygenase (ADO) variants. These pathways resulted in alkane production, with the best pathway achieving 38.3 mg L−1 heptadecane. We then engineered this strain to produce medium-chain alkanes by redirecting the elongation pathway through expression of phaG, and increased growth on xylose by overexpression of xylose reductase (XR). Therefore, we were able to achieve medium-chain alkane production on all three sugars of interest. Thus, these strains have potential as cell factories for oleochemicals from lignocellulose. In summary, this work highlights the potential of using systems biology, synthetic biology, and metabolic engineering to make D. hansenii CBS 767 a saltwater cell factory.