The best applicable expression host for an individual production process is typically determined based on various characteristics and attributes of the specific recombinant protein, like glycosylation pattern or post-translational modifications. Escherichia coli (E. coli) is by far the most widely used host organism for biopharmaceutical production of simple non-modified heterologous recombinant proteins with a low number of disulfide bonds. E. coli expression strains are favored due to their ability to quickly reach high cell densities in inexpensive media and their Food and Drug Administration (FDA)-approved status for human applications . Additionally, their well-characterized genetics and publicly available genome sequences [2–4], and the availability of a large number of cloning vectors and various mutant strains ensure that E. coli remains a valuable host for recombinant protein production . However, some drawbacks must be considered when selecting an E. coli expression system, including the limited ability to build disulfide bonds, the lack of an efficient secretion system due to the complexity of the two cell membranes, and the inability to perform posttranslational modifications that occur with eukaryotic proteins [6, 7].
Selection of an appropriate promoter is also a major step in process design. For high-level recombinant protein synthesis in E. coli, the promoter must be strong and inducible, and should exhibit a minimal level of basal transcriptional activity . The pET system (plasmid for expression by T7 RNA polymerase) applied in the present study meets most of these requirements [8, 9]; however, the resulting strong and high-level protein expression can overburden the metabolic capacities of the production host and lead to growth inhibition or even cessation after a short period of time [10, 11]. Therefore, the actual production phase during a process is very limited. Strong host/vector interactions also contribute to the stress level of the host, resulting in decreased recombinant protein yields far below the theoretical maximum. High-level protein production can also lead to modifications in catabolism and anabolism, including adjustments of the energy generating system, the protein producing system, and cellular fluxes [11, 12]. Overall, the cellular reactions caused by recombinant protein production are similar to the heat shock and stringent responses of E. coli . Altogether, these unfavorable conditions and the limitation of cellular resources caused by high production rates often result in accumulation of damaged and un- or mis-folded proteins in the host cell . Inclusion bodies (IB) sometimes form, depending on the specific folding behavior of the protein of interest; even small changes in the primary structure of a protein can influence its solubility .
Furthermore, in plasmid-based expression systems, the cellular energy required for plasmid maintenance and replication may influence the process outcome. In some cases the applied systems encounter problems with plasmid stability, leading to plasmid-free cell populations. To avoid plasmid-maintenance and stability problems, genome-based expression systems that integrate the target gene into the chromosome have been established for E. coli ; however, plasmid-based systems are still favored in industry because the cloning procedures are simpler and faster.
The present work focused on evaluating the prominent and very strong T7 expression system (commercialized by Novagen), within the most popular E. coli hosts, using an industrial-like process set-up. Previous comparisons of different strains have investigated biomass yield, growth behavior, or mRNA expression profiles [16–19]. For recombinant processes the impact of recombinant protein production onto the host metabolism is of also of great interest, as the host-vector interactions during such harsh conditions are often complex and unpredictable. Here we transformed the B strain BL21(DE3) and the two K–12 strains HMS174(DE3) and RV308(DE3) with the pET30a plasmid (Novagen; Germany) encoding the gene for human superoxide dismutase (SOD, EC 184.108.40.206) , and we evaluated their performances in fully induced fed-batch cultivations. The two K–12 hosts, RV308 a derivative of MG1655 and HMS174 of W3110, were compared to the well-known BL21 strain. All three are industrially relevant strains frequently used in large scale production processes . The model protein human superoxide dismutase used in this study can be expressed in soluble form in the cytoplasm of E. coli. However, strong expression systems and high amounts of inducer (IPTG) can lead to the accumulation of both soluble and insoluble SOD (inclusion bodies; IBs). Therefore, this protein was selected as model protein and the distribution of the soluble versus the insoluble protein form was monitored and used as quality criteria. We characterized the studied strains in terms of growth behavior, productivity, product quality (solubility), and overall system stability. The experiments were designed to provide valuable information about E. coli performance in industrial-like processes, which can be used in future process set-up and design.