Engineered E. coli keeps two plasmids alive without antibiotics or resistance genes

One of the quiet workhorses of modern biotechnology is also one of its most awkward compromises. Nearly every engineered strain of Escherichia coli used to make medicines, enzymes or fine chemicals carries its synthetic genes on plasmids, small circles of DNA that replicate independently of the chromosome. To stop cells from discarding these costly genetic passengers, manufacturers almost universally rely on antibiotics and, crucially, on the antibiotic resistance genes (ARGs) encoded on the plasmids themselves. A team at Gen-H Genetic Engineering Heidelberg GmbH has now shown that this dependence can be broken even for the demanding case of two plasmids maintained simultaneously, using a strategy they call conditional complementation. The work, published in Frontiers in Industrial Microbiology, demonstrates antibiotic-free selection and stable maintenance of two plasmids in ordinary, unsupplemented growth media, with productivity equal to or better than conventional antibiotic-based systems.

The sustainability argument against antibiotics in fermentation is straightforward. Industrial-scale antibiotic use adds cost, complicates regulatory approval, and keeps resistance genes circulating in production environments. A recent study cited by the authors found that standard biological waste decontamination methods do not fully inactivate DNA, underscoring the risk that plasmid-borne ARGs from synthetic strains could spread to environmental bacteria through horizontal gene transfer. Regulators are increasingly pressuring the industry to develop antibiotic-free biomanufacturing, and several commercial systems already exist for single-plasmid processes. The gap has been multi-plasmid applications: co-expression of chaperones to improve protein solubility, or metabolic engineering schemes that need plasmids with different copy numbers to balance enzyme expression. Existing dual-selection approaches based on two auxotrophic markers require repetitive genome engineering for every new strain-plasmid combination, while a CRISPRi-based genetic circuit, though an impressive demonstration of cellular programmability, involves many interacting components that could mutate and release the selection pressure.

The Heidelberg team’s approach builds on a system they described previously, in which an essential gene is removed from the cell’s normal transcriptional control and placed under an inducible promoter in the genome. The logic is elegant: if the essential gene is only switched on when a specific inducer sugar is added to the medium, then a plasmid-free cell can be kept alive simply by feeding it the inducer. But if the inducer is withheld, only cells that have taken up a plasmid carrying a working copy of the essential gene can grow. Transformation therefore becomes its own selection step, performed on plain agar with no antibiotic at all. Critically, because the plasmid-free strain remains viable under inducer supplementation, the same engineered host can be reused with different plasmids without repeating the genomic modification.

In their original implementation, the researchers used infA, the gene encoding Translation Initiation Factor 1, driven by the arabinose-inducible promoter. infA is an attractive marker: at just 219 base pairs it is tiny, it is essential under all culture conditions, and because its role is intracellular it eliminates cross-feeding, the phenomenon in which starving cells can be rescued by metabolites leaked from their neighbours. For the new study, the team first created an orthogonal variant by placing infA under the rhamnose promoter, which is controlled by the regulators RhaS and RhaR and is known, like the arabinose promoter, for its tight transcriptional regulation. The resulting strain grew robustly only when rhamnose was supplied. When transformed with a plasmid carrying a constitutively expressed infA cassette and plated on plain medium, every one of ten randomly picked colonies contained plasmid with the correct restriction digest pattern. Growth rates of the modified strain carrying the plasmid were statistically indistinguishable from the unmodified parental strain grown with antibiotic selection, showing that redirecting infA transcription to a plasmid imposes no measurable fitness cost.

To make the selection pressure visible, the researchers used a model plasmid encoding a chromoprotein built from two linked fluorescent proteins, E2-Crimson and mStrawberry, which together produce a purple pigment visible to the eye after extended incubation. After antibiotic-free transformation into the rhamnose-dependent strain, every colony on the plate developed purple pigmentation. The absence of pigment-free colonies demonstrated that only cells carrying the complementing plasmid had proliferated enough to form colonies, a striking visual confirmation that complementation alone provides genuine selection.

Plasmid maintenance over time was the next test, and here the design of the experiment matters. The team deliberately used a high-copy plasmid carrying a constitutively expressed fluorescent marker to impose a metabolic load, because uninduced plasmids are often stably maintained even without selection, and because the burden of expressing a productive gene is precisely what drives plasmid loss in real processes. After 72 hours of growth in plain LB medium, the unmodified strain showed a significant fraction of cells that had lost the plasmid, visible as reduced growth on antibiotic plates and loss of pigmentation. The engineered strain, by contrast, showed comparable growth and uniform pigmentation on all test media, and every single isolated colony still contained plasmid. The same pattern held for a second, entirely new system built around a different essential gene, secY, the central subunit of the bacterial Sec translocase that transports proteins across the inner membrane. secY posed a tougher engineering challenge because it sits in a densely packed chromosomal neighbourhood flanked by essential genes, so the team inserted an ectopic arabinose-inducible copy at the recAX locus and deleted the endogenous gene. An interesting biological detail supports this choice: in the absence of its partner SecE, SecY is rapidly degraded by the protease FtsH with a half-life of roughly two minutes, meaning any excess marker protein is efficiently recycled rather than accumulating.

With two independently validated complementation markers in hand, the team combined them. They built one strain in which both infA and secY are controlled by the arabinose promoter, and a second, more flexible strain in which infA is driven by the rhamnose promoter and secY by the arabinose promoter. In this dual-promoter strain, the two essential genes can be induced independently, which means the host can be propagated with one plasmid, two plasmids, or none, depending only on which inducers are added. When both compatible model plasmids, one complementing infA and the other secY, were co-transformed into the dual-promoter strain on plain medium, growth occurred only under conditions that allowed expression of both essential genes, whether through plasmid complementation or inducer supplementation. Specific growth rates of the dual-complementation strains carrying two plasmids showed no statistically significant difference from the unmodified parental strain under antibiotic selection.

The stability results were decisive. After 48 hours of growth in unsupplemented medium with two plasmids aboard, roughly only one third of colonies from the unmodified strain still carried both plasmids, with the high-copy plasmid lost particularly rapidly under the metabolic burden of constitutive fluorescent protein expression. In both engineered strains, every colony examined retained both plasmids, confirmed by growth on the corresponding antibiotic plates used purely as an analytical readout. The team then pushed the system into a realistic expression scenario using a two-plasmid T7 setup, in which one plasmid supplies T7 RNA polymerase from a lac promoter and the other expresses a fluorescent reporter from a T7 promoter, so that both plasmids must be present for any product to appear. T7 expression is notorious for destabilising plasmids because of its heavy metabolic burden. After five hours of induction, productivity under complementation matched that under antibiotic selection. After 24 hours, both engineered strains significantly outperformed the antibiotic-selected conventional strain, which lost productivity over time, likely because plasmid-encoded enzymes progressively inactivate the antibiotics and erode the selection pressure.

This durability points to a deeper advantage of the design. Antibiotic selection weakens as the drug is degraded, whereas complementation selects for cells expressing a beneficial gene, which is difficult to bypass. Simple inactivating mutations, the Achilles heel of toxin-antitoxin systems, accomplish nothing here; a cell can only escape selection by integrating the plasmid into the genome or by acquiring mutations that render the inducible promoter constitutively active, events the authors did not observe. The small size of infA makes recombination unlikely, and codon diversification between the genomic and plasmid copies could reduce homology further if needed. There is also a metabolic argument: ARGs are often expressed at unnecessarily high levels, and removing them frees cellular resources that can be channelled into the product. The authors note that the strategy previously worked across a wide range of E. coli strain backgrounds, including K-12 and B strains and more phylogenetically diverse isolates, and they expect the rhamnose and secY variants to transfer similarly, potentially even to other production organisms such as Bacillus, Corynebacterium or Pseudomonas where sufficiently tight inducible promoters and non-metabolic essential genes are available. By extending conditional complementation to two plasmids, and plausibly beyond with additional essential genes, the study offers biomanufacturing a route to cleaner, greener fermentation that asks nothing extra of the people doing the work: the handling, the transformations and the media remain exactly as convenient as before, only without the antibiotics.

Subject of Research: Antibiotic-free maintenance of multiple plasmids in E. coli using conditional complementation of essential genes

Article Title: Antibiotic-free maintenance of multiple plasmids in Escherichia coli using conditional complementation

Article References: Brechun, K. E., Förschle, M., Schmidt, M., & Kranz, H. (2026). Antibiotic-free maintenance of multiple plasmids in Escherichia coli using conditional complementation. Frontiers in Industrial Microbiology, 4, Article 1913754. https://doi.org/10.3389/finmi.2026.1913754

Image Credits: AI Generated

DOI: 10.3389/finmi.2026.1913754

Keywords: antibiotic-free selection, Escherichia coli, plasmid maintenance, conditional complementation, genome engineering, biomanufacturing, infA, secY, inducible promoters, antimicrobial resistance, precision fermentation, synthetic biology