Knowledge

Making a Protein in E. coli: Host, Gene Design and the Limits on Yield

Expressing a recombinant protein in Escherichia coli — a bacterium used as a production host — means putting a gene that the cell does not normally carry (a heterologous gene) onto a plasmid, switching it on, and letting the culture build the protein. It is the first step that fixes almost everything downstream: how much protein there is, whether it is folded, where in the cell it ends up, and what has to be removed from it later. The published work on this step is consistent on one point: the arrangement that produces the most messenger RNA is often not the arrangement that produces the most usable protein.

Why E. coli is the default host

The low cost and simplicity of cultivating bacteria make the E. coli expression system a preferable choice for producing therapeutic proteins, both at lab scale and in industry [3]. Straightforward recombinant DNA technology also offers engineering tools to produce protein molecules with modified features [3]. A 2023 review in Archives of microbiology describes E. coli as one of the popular hosts for recombinant proteins, and notes that scientists have reported bottlenecks in this host that create a need to improve it [6]. A 2022 review in Microbial cell factories puts the limitation more sharply: conventional expression strains are unable to effectively express proteins with complex structures or toxicity [7].

The host also lacks machinery that eukaryotic cells have. Bacterial cells have no post-translational modification mechanisms such as glycosylation or proteolytic protein maturation, and only a limited capacity for forming disulfide bridges [3]. That gap can, to a certain extent, be overcome with protein engineering, which is also often used to improve protein stability [3]. More elaborate designs are possible through genetic fusions of two proteins [3]. The same review emphasises the importance of designing a construct without any unnecessary amino acid residues [3] — a point that matters for any protein where the finished molecule must have a defined sequence and nothing more.

The promoter decides more than transcript level

A systematic comparison published in Microbial cell factories standardised the plasmid backbones so that regulator/promoter systems could be compared on equal terms, using the RK2 and pMB1 origins of replication and the XylS/Pm wild type, a Pm variant called XylS/Pm ML1-17, LacI/PT7lac, LacI/Ptrc and AraC/PBAD systems [4]. High expression levels correlated with high replicon copy number, and the LacI/PT7lac system generated more transcript than all four of the others [4].

That transcriptional advantage did not translate directly into protein. The authors report that more transcript does not always lead to correspondingly more efficient protein production, particularly if protein functionality is considered [4]. In most cases the XylS/Pm ML1-17 and LacI/PT7lac systems gave the highest amounts of functional protein, and XylS/Pm ML1-17 was the most flexible because it does not require any specific features of the host [4]. The AraC/PBAD system was very good with respect to tightness — how completely it stays off before induction — and a bioinformatics prediction tool suggested it has the most translation-efficient untranslated region [4]. Flow cytometry of individual cells indicated that cell-to-cell heterogeneity is very relevant to understanding protein production at the population level [4].

The conclusion the authors draw is that the choice of expression system needs to be evaluated for each specific case, and that standardised vectors make it easier to identify the nature of case-specific bottlenecks [4]. In other words, outcomes of process development for a specific protein remain often unpredictable, partly because expression cassettes behave differently in different genetic contexts [4].

Secretion to the periplasm has a hard capacity limit

To send a protein there, it is fitted with an N-terminal signal sequence so it can cross the cytoplasmic membrane through the protein-conducting Sec-translocon [1]. Production of secretory heterologous proteins is often toxic to the cell, which limits yields [1].

A study using the Lemo21(DE3) strain, in which the expression intensity of the target gene can be precisely controlled, produced two proteins — super folder green fluorescent protein and a single-chain variable antibody fragment — each carrying a DsbA-derived signal sequence [1]. Running the same construct at different expression levels let the authors find an optimal level for each gene [1]. Too high an expression level saturated the Sec-translocon capacity: translocation of the cell's own secretory proteins was hampered, and misfolding and aggregation appeared in the cytoplasm [1]. At the optimal level, those negative effects were minimised and periplasmic yields were optimised [1]. The authors conclude that saturating the Sec-translocon is a major bottleneck, and that it can be relieved by harmonising the expression level of the heterologous gene with the translocon's capacity [1].

Pushing harder costs the cell something

Redirecting a cell's metabolism towards a product can leave it with stress symptoms: a decreased growth rate, impaired protein synthesis, genetic instability and an aberrant cell size [8]. On an industrial scale this shows up as processes that are not economically viable [8]. Most of these symptoms are attributed in the literature to "metabolic burden", but the actual triggers and stress mechanisms are poorly understood [8]. A 2024 review concludes that activation of the different stress mechanisms is complex and that many are interconnected, and that care has to be taken when overexpressing heterologous proteins because the cell's metabolism is tightly regulated [8]. The 2022 review frames the same problem as an optimisation target: the key is to reduce the host burden associated with protein overproduction while improving the cell's ability to fold and modify proteins accurately at high expression levels [7].

Tuning more than one element at a time

Where several genes have to work together, single-variable tuning is not enough. Work published in Nature communications on a 15-step heterologous astaxanthin route describes multivariate modular methods, which reduce regulatory complexity by grouping genes into modules but balance between modules better than within them [2]. The authors developed a multidimensional heuristic process that balances modules by varying promoter strength while coordinating activity inside each module using ribosome binding sites and enzyme variants [2]. Applied to astaxanthin, it raised production to 184 mg l⁻¹ day⁻¹, or 320 mg l⁻¹, and it also improved yields of nerolidol and linalool [2].

Whole gene clusters can also be moved into E. coli. A minimal nitrogen fixation cluster of nine genes from a Paenibacillus strain, organised as a single operon, enabled synthesis of catalytically active nitrogenase in E. coli when expressed either from the native nifB promoter or from the T7 promoter [5]. Notably, transcription from the nifB promoter was constitutive in E. coli, meaning the negative regulation present in the original organism was bypassed in the heterologous host [5]. That is a useful warning: a control element that behaves one way in its native cell may behave differently once transferred.

What to take from this when choosing a process

Ask what is being measured. A high transcript level or a strong band is not the same as functional protein, and the comparison study found the two can diverge [4]. Ask where the protein is being made. If it is directed to the periplasm, the expression level should be matched to translocon capacity rather than maximised, and the evidence for that match should be data at several expression levels, not a single run [1]. Ask what the construct contains: a design with no unnecessary amino acid residues is the stated goal [3]. Ask whether the strain is paying for the yield — decreased growth rate, genetic instability and impaired protein synthesis are recognised signs of metabolic burden, and they are what make a process uneconomic at scale [8].

References

  1. Schlegel S, Rujas E, Ytterberg AJ, Zubarev RA, Luirink J, de Gier JW. Optimizing heterologous protein production in the periplasm of E. coli by regulating gene expression levels — Microbial cell factories, 2013-03-12.
  2. Zhang C, Seow VY, Chen X, Too HP. Multidimensional heuristic process for high-yield production of astaxanthin and fragrance molecules in Escherichia coli — Nature communications, 2018-05-11.
  3. Kamionka M. Engineering of therapeutic proteins production in Escherichia coli — Current pharmaceutical biotechnology, 2011-02-01.
  4. Balzer S, Kucharova V, Megerle J, Lale R, Brautaset T, Valla S. A comparative analysis of the properties of regulated promoter systems commonly used for recombinant gene expression in Escherichia coli — Microbial cell factories, 2013-03-18.
  5. Wang L, Zhang L, Liu Z, Zhao D, Liu X, Zhang B, Xie J, Hong Y, Li P, Chen S, Dixon R, Li J. A minimal nitrogen fixation gene cluster from Paenibacillus sp. WLY78 enables expression of active nitrogenase in Escherichia coli — PLoS genetics, 2013-10-17.
  6. Pouresmaeil M, Azizi-Dargahlou S. Factors involved in heterologous expression of proteins in E. coli host — Archives of microbiology, 2023-04-29.
  7. Zhang ZX, Nong FT, Wang YZ, Yan CX, Gu Y, Song P, Sun XM. Strategies for efficient production of recombinant proteins in Escherichia coli: alleviating the host burden and enhancing protein activity — Microbial cell factories, 2022-09-15.
  8. Snoeck S, Guidi C, De Mey M. "Metabolic burden" explained: stress symptoms and its related responses induced by (over)expression of (heterologous) proteins in Escherichia coli — Microbial cell factories, 2024-03-30.

All articles