The impact of oxygen on the transcriptome of recombinant S. cerevisiae and P. pastoris - a comparative analysis

Abstract

Background: Saccharomyces cerevisiae and Pichia pastoris are two of the most relevant microbial eukaryotic platforms for the production of recombinant proteins. Their known genome sequences enabled several transcriptomic profiling studies under many different environmental conditions, thus mimicking not only perturbations and adaptations which occur in their natural surroundings, but also in industrial processes. Notably, the majority of such transcriptome analyses were performed using non-engineered strains.In this comparative study, the gene expression profiles of S. cerevisiae and P. pastoris, a Crabtree positive and Crabtree negative yeast, respectively, were analyzed for three different oxygenation conditions (normoxic, oxygen-limited and hypoxic) under recombinant protein producing conditions in chemostat cultivations.

Results: The major differences in the transcriptomes of S. cerevisiae and P. pastoris were observed between hypoxic and normoxic conditions, where the availability of oxygen strongly affected ergosterol biosynthesis, central carbon metabolism and stress responses, particularly the unfolded protein response. Steady state conditions under low oxygen set-points seemed to perturb the transcriptome of S. cerevisiae to a much lesser extent than the one of P. pastoris, reflecting the major tolerance of the baker's yeast towards oxygen limitation, and a higher fermentative capacity. Further important differences were related to Fab production, which was not significantly affected by oxygen availability in S. cerevisiae, while a clear productivity increase had been previously reported for hypoxically grown P. pastoris.

Conclusions: The effect of three different levels of oxygen availability on the physiology of P. pastoris and S. cerevisiae revealed a very distinct remodelling of the transcriptional program, leading to novel insights into the different adaptive responses of Crabtree negative and positive yeasts to oxygen availability. Moreover, the application of such comparative genomic studies to recombinant hosts grown in different environments might lead to the identification of key factors for efficient protein production.

Figure 1

Global comparison of differential gene regulation. The correlation matrix of 2191 common genes (absolute normalized expressions levels) for (A) S. cerevisiae and (B) P. pastoris strains grown under normoxic (N), oxygen-limited (L) and hypoxic (H) conditions are visualized with the principle component analysis (PCA) biplot. Lines pointing in the same direction correspond to strains and oxygen conditions which are correlated. In both strains, the first and second components explain 97 % of the total variation. Overlapping sets of regulated genes in (C) S. cerevisiae, (D) P. pastoris and (E) between both species are displayed by Venn diagrams. Fab = Fab producing strain, Cont = control strain. HvsL = hypoxic vs. oxygen-limited, HvsN = hypoxic vs. normoxic, and LvsN = oxygen-limited vs. normoxic.

Figure 2

GO group enrichment as determined by Fisher's exact test. Significantly enriched GO functional groups (computed with the Fisher's exact test with a p-value ≤ 0.05) are labeled in red (upregulated genes), green (downregulated genes) and orange (both up- and downregulated genes). Pairwise comparisons in S. cerevisiae (Sc) and P. pastoris (Pp) are abbreviated as in figure 1.

Figure 3

Overlay of transcriptome data on the S. cerevisiae metabolic map. Fold change data of the pairwise comparison hypoxic vs. normoxic (HvsN) of the recombinant S. cerevisiae (A) and P. pastoris (B) strain are overlapped with the metabolic map of S. cerevisiae (MetaCyc, SGD database [66]). Each node in the diagram represents a single metabolite, and each line represents a single bioreaction. In the right part of the diagram the small molecule metabolism is represented (for a complete description of the map see http://pathway.yeastgenome.org). Reaction lines are colour-coded (three colour bins) according to the fold change value of the gene: red for data values that exceed a log2 fold change threshold of 0.59, yellow for data values less than the inverse of the threshold, and blue for values in between. Detailed lists of all the regulated pathways, together with their diagrams and corresponding gene lists are provided in additional file 2 (for S. cerevisiae) and 3 (for P. pastoris). Depicted pathways are indicated by numbers, according to the table shown in additional file 4. The ergosterol pathway (n.20) and glycolysis (n.69) are indicated by dashed boxes.

Figure 4

Clustering comparison graphical output. The output of the hierarchical clustering (correlation based distance, average linkage) of the fold changes of all pairwise comparisons of oxygen provision is displayed in form of a dendrogram attached to a heat map, with the two main branches indicated by a red line (A). The clustered matrix is linked to the outcome (line-plots) of the k-means clustering (k = 10) (B). The clustering comparison correspondence is displayed in the central part of the graph. It is depicted as lines of varying thickness, mapping sub-branches of the tree to flat clustering superclusters. Line thickness is proportional to the number of elements common to both sets. Pairwise comparisons in S. cerevisiae (Sc) and P. pastoris (Pp) are abbreviated as follows: HvsL (hypoxic vs. oxygen-limited), HvsN (hypoxic vs. normoxic) and LvsN (oxygen-limited vs. normoxic). Fab = Fab producing strain, Cont = control strain.