RNA-seq reveals multifaceted gene expression response to Fab production in Escherichia coli fed-batch processes with particular focus on ribosome stalling

Abstract

Background: Escherichia coli is a cost-effective expression system for production of antibody fragments like Fabs. Various yield improvement strategies have been applied, however, Fabs remain challenging to produce. This study aimed to characterize the gene expression response of commonly used E. coli strains BL21(DE3) and HMS174(DE3) to periplasmic Fab expression using RNA sequencing (RNA-seq). Two Fabs, Fabx and FTN2, fused to a post-translational translocation signal sequence, were produced in carbon-limited fed-batch cultivations.

Results: Production of Fabx impeded cell growth substantially stronger than FTN2 and yields of both Fabs differed considerably. The most noticeable, common changes in Fab-producing cells suggested by our RNA-seq data concern the cell envelope. The Cpx and Psp stress responses, both connected to inner membrane integrity, were activated, presumably by recombinant protein aggregation and impairment of the Sec translocon. The data additionally suggest changes in lipopolysaccharide synthesis, adjustment of membrane permeability, and peptidoglycan maturation and remodeling. Moreover, all Fab-producing strains showed depletion of Mg²⁺, indicated by activation of the PhoQP two-component signal transduction system during the early stage and sulfur and phosphate starvation during the later stage of the process. Furthermore, our data revealed ribosome stalling, caused by the Fabx amino acid sequence, as a contributor to low Fabx yields. Increased Fabx yields were obtained by a site-specific amino acid exchange replacing the stalling sequence. Contrary to expectations, cell growth was not impacted by presence or removal of the stalling sequence. Considering ribosome rescue is a conserved mechanism, the substantial differences observed in gene expression between BL21(DE3) and HMS174(DE3) in response to ribosome stalling on the recombinant mRNA were surprising.

Conclusions: Through characterization of the gene expression response to Fab production under industrially relevant cultivation conditions, we identified potential cell engineering targets. Thereby, we hope to enable rational approaches to improve cell fitness and Fab yields. Furthermore, we highlight ribosome stalling caused by the amino acid sequence of the recombinant protein as a possible challenge during recombinant protein production.

Keywords: Envelope stress; Periplasmic expression; Polyproline; Recombinant protein production; Ribosome stalling; Transcriptomics.

Fig. 1

Final cell dry mass A, and intra- and extracellular soluble Fab yields B produced in different recombinant E. coli expression systems in fed-batch processes after 16 h of induction. Samples were analyzed in biological triplicates (n = 3, error bars represent standard error of the mean (SEM)). CDM of Fab-producing strains was compared to the respective wild-type strains BL21(DE3) and HMS174(DE3) (* … p < 0.05, ** … p < 0.01, *** p < 0.001, two sample t-test). The soluble Fab yields in the intra- and extracellular fractions were determined using ELISA. Fab expression patterns were analyzed in LC-specific WBs C and the ratio of LC to Fab in the soluble and IB fractions was estimated D

Fig. 2

Number of up- and downregulated genes in recombinant E. coli BL21(DE3)- (A) and HMS174(DE3)-based expression systems (B) after 2 and 12 h of Fab production relative to the non-induced cells in fed-batch processes. Differential gene expression was analyzed using DESeq2 from biological triplicates (n = 3)

Fig. 3

Venn diagrams of differentially expressed genes in Fab-producing E. coli expression systems 2 and 12 h after induction. Gene expression was compared between Fabx and FTN2 producing expression systems for BL21(DE3) (A) and HMS174(DE3) (B). Gene expression was compared between the two Fabx-producing strains (C). Genes also differentially expressed in wild-type BL21(DE3) and HMS174(DE3) were excluded from the analysis

Fig. 4

Log2FC of genes involved in ribosome stalling and rescue in Fab-producing strains after 12 h of induction relative to the respective non-induced samples (A). Alignment of the region of the Fabx sequence that causes ribosome stalling to the FTN2 sequence (B). The ribosome stalling sequence, PPG, is located in the variable domain of the Fabx-HC

Fig. 5

Cell growth (A) and Fab yields (B) in fed-batch cultivations of recombinant BL21(DE3)-based expression systems producing Fabx and Fabx(P40A), with and without PPG in the HC sequence, respectively. Intra- and extracellular Fabx(P40A) content as determined by ELISA, was compared to Fabx at different timepoints (C). Cultivations were done in triplicates (n = 3, error bars represent SEM)