The effect of hypoxia on the lipidome of recombinant Pichia pastoris

Abstract

Background: Cultivation of recombinant Pichia pastoris (Komagataella sp.) under hypoxic conditions has a strong positive effect on specific productivity when the glycolytic GAP promoter is used for recombinant protein expression, mainly due to upregulation of glycolytic conditions. In addition, transcriptomic analyses of hypoxic P. pastoris pointed out important regulation of lipid metabolism and unfolded protein response (UPR). Notably, UPR that plays a role in the regulation of lipid metabolism, amino acid metabolism and protein secretion, was found to be upregulated under hypoxia.

Results: To improve our understanding of the interplay between lipid metabolism, UPR and protein secretion, the lipidome of a P. pastoris strain producing an antibody fragment was studied under hypoxic conditions. Furthermore, lipid composition analyses were combined with previously available transcriptomic datasets to further understand the impact of hypoxia on lipid metabolism. Chemostat cultures operated under glucose-limiting conditions under normoxic and hypoxic conditions were analyzed in terms of intra/extracellular product distribution and lipid composition. Integrated analysis of lipidome and transcriptome datasets allowed us to demonstrate an important remodeling of the lipid metabolism under limited oxygen availability. Additionally, cells with reduced amounts of ergosterol through fluconazole treatment were also included in the study to observe the impact on protein secretion and its lipid composition.

Conclusions: Our results show that cells adjust their membrane composition in response to oxygen limitation mainly by changing their sterol and sphingolipid composition. Although fluconazole treatment results a different lipidome profile than hypoxia, both conditions result in higher recombinant protein secretion levels.

Keywords: Antibody fragment; Hypoxia; Lipidomics; Pichia pastoris; Protein secretion; Recombinant protein production; Unfolded protein response.

Fig. 1

Schematic representation of lipid biosynthesis pathways from P. pastoris and its regulation in hypoxia. Genes under hypoxic conditions were compared to normoxic conditions. Lipid species analyzed in the study are boxed, and genes selected to perform transcriptional analysis by quantitative PCR (ddPCR) are underlined. Fold changes of genes are indicated by color: red upregulated genes, green downregulated genes, grey no significant changes. (Based on p values <0.05). Transcriptional data was taken from [23]

Fig. 2

Cellular fatty acid composition. Fatty acid composition (% of total) of P. pastoris cells producing the Fab 2F5 and growing under normoxic or hypoxic conditions in the presence or absence of fluconazole. Data represent mean values ± SD from triplicates. *p < 0.05 for the t tests

Fig. 3

Cellular phospholipid composition. Phospholipid composition (% of total phospholipids) of the cells growing under normoxic or hypoxic conditions, in the presence or absence of fluconazole. PC phosphatidylcholine, PA phosphatidic acid, PI phosphatidylinositol, PS phosphatidylserine, Lyso-PL lysophospholipids, PE phosphatidylethanolamine, CL cardiolipin, DMPE dimethyl phosphatidylethanolamine. Data represent mean values ± SD from duplicates. *p < 0.05 for the t tests comparing phospholipid detected values

Fig. 4

Sphingolipid composition. Sphingolipid analysis of cells growing under normoxic or hypoxic conditions in the presence or absence of fluconazole. Sphingolipid molecular species of ceramides (Cer), hexosylceramides (HexCer), inositolphosphorylceramides (IPC), mannosyl-inositolphosphorylceramides (MIPC) and mannosyl-diinositolphosphorylceramides (M(IP)₂C) are shown. Species are expressed as long-chain-base/fatty acyl. LCB and fatty acyls are expressed as number of carbons: number of C–C double bonds; number of hydroxyl groups. *p < 0.05 for the t tests

Fig. 5

Principal component analysis (PCA) of lipidomic data. Principal component analysis of the lipidome data in a biplot of components one and two. The biplot shows lipidomic data (scores) as labelled dots and treatment effect (loadings) as vectors. Vectors that are close together are highly correlated in terms of the observed lipidomic content, while vectors that are orthogonal are poorly correlated. PC1 correlates well with the change due to fluconazole treatment, whereas PC2 appears to be correlated with the change in oxygen conditions