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

A key sustainability concern in biotechnology is the heavy reliance on antibiotics and antibiotic resistance genes (ARGs) for plasmid selection and maintenance. This is leading to increasing regulatory pressure to develop antibiotic-free biomanufacturing processes. Currently, there are limited options enabling antibiotic-free use of multiple plasmids, which is often required in processes with advanced genetic complexity. Here, we use conditional complementation of two essential genes in Escherichia coli to enable ARG-free selection and maintenance of two plasmids in non-supplemented media. First, we develop orthogonal strains containing an essential gene under the control of an inducible promoter; as a result, the growth of each strain is dependent on the corresponding inducer or the presence of a complementing plasmid. We characterize these strains, demonstrating straightforward, antibiotic-free plasmid transformations and robust plasmid maintenance in non-supplemented media. Next, by combining these modifications, we create strains enabling maintenance of two plasmids in non-supplemented media without requiring ARGs. We demonstrate that by using two orthogonal inducible promoters, it is even possible to create a strain where one or two complementing plasmids can be used. By expressing model genes of interest, we show the productivity of these strains is equivalent or superior to conventional, antibiotic-dependent strains under selective conditions. This method offers a robust, antibiotic-free solution for multi-plasmid biosynthetic pathways, enabling improved sustainability without imposing any technical restrictions or compromising handling.

White biotechnology is well-positioned to offer sustainable solutions to produce high-value-added products from renewable resources under environmentally friendly conditions. With improvements in biotechnology tools and techniques, the list of molecules that can be manufactured through precision fermentation of engineered cell factories is ever increasing. However, a key challenge in biotechnology remains the widespread dependence on antibiotics and antibiotic resistance genes (ARGs) for plasmid selection and maintenance. Replacing antibiotic supplementation with antibiotic-free plasmid maintenance technology would make biomanufacturing cleaner, greener and more sustainable, with the added benefits of cost reduction and simplified regulatory compliance.

Numerous strategies have been developed that offer in vivo antibiotic-free plasmid maintenance (Sodoyer et al., 2012). These systems use neutralization of a negative selectable marker (Szpirer and Milinkovitch, 2005; Gabant, 2013) and/or complementation (Fiedler and Skerra, 2001; Vidal et al., 2008; Amrofell et al., 2023) to ensure plasmid maintenance. Some of these systems have moved beyond the proof-of-concept phase, and are being used in industry (Luke et al., 2009; Reschner et al., 2013; Brechun et al., 2026).

Some biomanufacturing processes rely on multiple plasmids, e.g., to co-express chaperones for improved solubility, or for use in metabolic engineering, where plasmids with different copy numbers may be required to balance enzyme expression (Piao et al., 2019; Wang et al., 2022; Taylor et al., 2023). However, there are few systems available enabling antibiotic-free maintenance of multiple plasmids. Dual complementation using two auxotrophic markers has been demonstrated (Schneider et al., 2005; Kang et al., 2018); however, constructing new strain-plasmid combinations requires repetitive genome engineering, making the strategy cumbersome. Recently, a genetic circuit was developed to maintain two plasmids antibiotic-free using a CRISPRi-based inverter circuit (Kim et al., 2022). While an impressive demonstration of cellular programmability, the numerous interacting components may carry an elevated risk for the loss of selection pressure through spontaneous mutation.

We recently developed a plasmid maintenance system using conditional complementation of an essential gene, which offers ARG-free complementation with convenient handling. Briefly, the host strain genome was modified by placing an essential gene under the control of an inducible promoter; as a result, strain growth is dependent on inducer supplementation or on the presence of a complementing plasmid (Brechun et al., 2024, 2026). Critically, the ability to propagate the plasmid-free engineered strain allows transformations to be performed without requiring an ARG, simply by plating the transformed cells on media without the inducer. Therefore, a genetically modified strain can be reused with different plasmids without needing to repeat the genomic modification. We sought to apply this principle to enable convenient antibiotic-free plasmid maintenance of two complementing plasmids.

In our original system, we used the essential gene infA, which encodes Translation Initiation Factor 1 (IF1), and the arabinose inducible promoter (P Ara), allowing for infA complementation or arabinose (Ara)-based plasmid-free strain propagation. The gene infA is small (219 base pairs), essential under all culture conditions, has an intracellular role (eliminating cross-feeding (Cummings and Hershey, 1994)) and is known to be well-suited for use as a selectable marker in Δ infA strains (Isaksson et al., 2002; Kang et al., 2018; Pagni et al., 2022; Sathesh-Prabu et al., 2022; Park et al., 2024; Outram et al., 2025; Yiakoumetti et al., 2025). The P Ara is well known for its tight transcriptional control (Guzman et al., 1995).

To create an orthogonal system, we first developed a strain using the rhamnose promoter (P Rha) to drive genomic expression of infA. P Rha is controlled by the regulators RhaS and RhaR, which activate transcription in the presence of the inducer, rhamnose (Rha) (Wickstrum et al., 2005). Similar to P Ara, P Rha is known for its tight transcriptional regulation (Giacalone et al., 2006). We show that when infA is under the control of P Rha, the growth of the resulting strain is dependent on the presence of Rha or a complementing plasmid, allowing for antibiotic-free transformations and stable, antibiotic-free plasmid-maintenance.

Next, we designed an analogous system using a different essential gene, namely secY. The SecY protein is the central subunit of the bacterial Sec translocase, which transports proteins from the cytoplasm to the periplasm or into the inner membrane (Lycklama a Nijeholt and Driessen, 2012). SecY binds with SecE to form a membrane channel. Interestingly, it has been shown that in the absence of SecE, SecY is rapidly degraded by FtsH with a half-life of approximately 2 minutes (Taura et al., 1993; Kihara et al., 1995). Due to its intracellular role, complementation with secY is expected to yield in non-feedable selection. Furthermore, the rapid degradation of unbound SecY should ensure that the excess ‘marker protein’ is recycled. In comparison to infA, the secY coding sequence is longer (1332 bp); however, the sequence is only slightly longer than an antibiotic resistance marker (e.g., tetracycline 1191 bp). We engineered a strain with Ara-inducible genomic secY expression and show that the growth of the strain is dependent on the presence of Ara or a complementing plasmid. We use the new strain to demonstrate antibiotic-free transformation and stable plasmid maintenance via secY complementation.

We then combined these components to demonstrate conditional complementation to enable antibiotic-free plasmid maintenance of two plasmids with convenient handling. First, we created a strain where two essential genes (infA and secY) are each under the control of P Ara, allowing dual plasmid complementation or strain propagation via Ara supplementation. Following this, we created a strain where the essential genes are under the control of two different promoters, namely with infA expression being driven by P Rha and secY expression being driven by P Ara. This allows dual plasmid complementation, or strain propagation via Ara and Rha supplementation. In addition, the use of two orthogonal promoters allows the strain to be propagated with a single plasmid if necessary, through supplementation with the corresponding inducer in the absence of the complementing plasmid. We show that growth of these new strains is dependent on the presence of complementing plasmid(s) and/or inducer(s), and that this strategy enables antibiotic-free transformation and robust plasmid maintenance. Finally, we demonstrate antibiotic-free protein expression from two plasmids, with expression levels comparable to those achieved in a conventional, antibiotic-based expression. Overall, these modifications highlight the robust nature of the conditional complementation strategy, and allow for excellent experimental flexibility in antibiotic-free plasmid use.

Bacterial growth conditions

Cloning and strain engineering was performed in E. coli HS996, a DH10β derivative with a mutation in the fhuA gene. The unmodified reference strains and the modified strains with plasmid(s) were grown in LB. In the absence of a complementing plasmid, the modified strains were grown in LB supplemented with 0.25% Ara and/or 0.25% Rha. To evaluate growth of the modified strains on different media, an overnight culture was grown in LB (with inducer supplementation in the absence of a complementing plasmid), diluted to an optical density at 600 nm (OD 600) of 0.02 in LB, and then streaked on LB plates with inducer supplementation as indicated. When indicated, the following antibiotics were used: chloramphenicol (Cm, 30 µg/ml for high-copy plasmids and 15 µg/ml for low-copy plasmids and genomic expression), Zeocin™ (Zeo, 50 µg/ml for high-copy plasmids and 25 µg/ml for low-copy plasmids and genomic expression), kanamycin (Kan, 15 µg/ml), tetracycline (Tet, 3 µg/ml) and ampicillin (Amp, 100 µg/ml). Handling of bacterial cultures in the absence of antibiotic selection was performed in an MSC-Advantage™ Class II Biological Safety Cabinet (Thermo Fisher Scientific) and growth was accompanied with an uninoculated medium control to monitor for contamination.

Strain engineering

Strain modifications were performed using λ-Red recombineering and recombinase-mediated marker deletion, as described previously (Wang et al., 2006). All strains used and developed in this study are outlined in Table 1. Detailed genotypes are shown in the SI (Supplementary Table 1). In brief, genome modifications were performed using a low-copy, temperature-sensitive helper plasmid expressing the λ-Red genes under the control of P Ara or P Rha. Genome integration was performed by electroporating 200–1000 ng of linear DNA with 50–200 bp of flanking homologous sequence into induced, electrocompetent cells using an Eppendorf Electroporator 2510 (1350 V). Fragments for integration were prepared via PCR, digested with DpnI and purified with a GeneJET PCR Purification kit (Thermo Fisher Scientific). Marker deletion was achieved using Flp and Cre recombinases, expressed from low-copy, temperature-sensitive helper plasmids with temperature-inducible expression, as described previously (Brechun et al., 2024). Marker deletion was confirmed by patching isolated colonies on LB with and without the antibiotic targeted for deletion (with inducer supplementation as required), and antibiotic-sensitive colonies were selected for further analysis. All modifications were confirmed using Sanger sequencing (Microsynth AG).

Plasmid generation

All plasmids used in this study are described in Table 2. Plasmid maps and sequences are provided in the SI. Plasmids were assembled using λ-Red recombineering or SLiCE (Zhang et al., 2012). To assemble CGB273, araC, P Ara and bla (Amp R) in CGB002 was replaced by a fragment containing rhaRS, P Rha and tetA using λ-Red recombineering. To assemble CGB326, cat (Cm R) and infA in CGB237 were replaced by a fragment containing zeo and secY using λ-Red recombineering. To assemble CGB332, cat (Cm R) and infA in CGB285 were replaced by a fragment containing zeo and secY using λ-Red recombineering. To assemble CGB327, a fragment containing infA, cat (Cm R) and the ColE1 origin from CGB285 and the mStrawberry coding sequence from CGB285 were assembled using SLiCE. To assemble CGB328, a fragment containing the E2-Crimson coding sequence and a fragment containing bla and the p15A origin (from pACYC177, New England Biolabs) were assembled using SLiCE. Next, bla was replaced by a fragment containing zeo and secY using λ-Red recombineering. To create CGB334, the E2-Crimson coding sequence and zeo in CGB328 were replaced with a fragment containing aph(3’)-II (Kan R). To assemble CGB342, a PCR fragment containing secY, zeo, the p15A origin and the lac promoter (P lac) was generated using CGB328 as a template and assembled using SLiCE with a PCR fragment containing the T7 RNA polymerase gene, amplified from the BL21(DE3) genome. All modifications were confirmed using restriction digest analysis and sequencing (Sanger and/or nanopore, Microsynth AG).

Plasmid transformation and analysis

Each plasmid used in this study contains an ARG and a complementation marker for use in the unmodified and modified strains respectively. This ensured the activity of both strains could be compared with exactly the same plasmids, and provided a convenient form of selection to assess for plasmid presence at the end of an experiment. In the modified strains, plasmid selection and maintenance was achieved solely via complementation.

The antibiotic-free transformations were performed using a standard electroporation protocol with the exception that during generation of the electrocompetent cells, the cultures were grown with inducer supplementation (Ara and/or Rha). Cultures were electroporated using cooled cuvettes (1 mm gap) in an Eppendorf Electroporator 2510 set to 1.35 kV, 10 F, 600Ω. After electroporation, the cells were incubated in 1 ml of LB at 37 °C with agitation for 1 h, then plated on LB-agar and incubated overnight at 37 °C. We found transformation using complementation yielded significantly more colonies, so we reduced the volume of culture plated to obtain isolated colonies. To analyze plasmid content after the transformation, isolated colonies were picked, inoculated into 1.4 ml of LB media and grown overnight at 37 °C in an Eppendorf Thermomixer with agitation. Plasmid was isolated using a GeneJET Plasmid Miniprep kit (Thermo Fisher Scientific) and analyzed using restriction enzymes.

Plasmid maintenance

To examine plasmid maintenance, a single colony was inoculated into 1.4 ml of non-supplemented LB and incubated for 2 days (two plasmids) or 3 days (single plasmid) at 37 °C with agitation and passaging every 24 h. In each case, the volume of the final overnight culture was increased to 50 ml to yield sufficient culture for downstream analysis. Each final culture was streaked on an LB plate to isolate single colonies, and the isolated colonies were patched on LB and LB with antibiotics (Zeo or Cm) to analyze for plasmid presence in individual cells. In parallel, an aliquot of the final culture was adjusted to an OD 600 of 2.5 and used to create a dilution series in LB. Next, 2.5 µl of each dilution were plated on LB, LB with inducer (Ara or Rha), and LB with antibiotics (Zeo or Cm) as indicated. The plates were incubated overnight at 37 °C. Prior to imaging, the plates were further incubated at 4 °C to facilitate development of the chromoprotein pigment. Plate images were optimized through a global adjustment of the brightness and contrast.

Microtiter plate analysis

Bacterial growth rates and protein expression were monitored using the multimode microplate reader Varioskan™ ALF (Thermo Fisher Scientific) and were analyzed using SkanIt software (Thermo Fisher Scientific). For the growth curves of the modified strains without a plasmid, single colonies were inoculated into 1.4 ml of LB supplemented with the appropriate inducer and grown overnight at 37 °C in an Eppendorf Thermomixer with agitation. The following morning, the cultures were diluted to an OD 600 of 0.005 in LB, and then 20 µl were inoculated into each 180 µl of LB and 180 µl of LB with inducer in a flat-bottom, clear 96-well microtiter plate. The plate was sealed with a Breathe-Easy sealing membrane (Merck KGaA) and incubated at 37 °C with non-continuous orbital shaking (5 s on, 10 s off) for 24 h with OD 600 measurements every 10 min. For the growth curves, strains were transformed with an appropriate low-metabolic load plasmid (using non-supplemented LB and antibiotics for selection in modified and unmodified strains respectively). Precultures were grown overnight in 1.4 ml of LB (with antibiotics for the unmodified strains) at 37 °C in an Eppendorf Thermomixer with agitation. The cultures were diluted 1:100 in LB (with antibiotics for the unmodified strains) with a total volume of 200 µl in flat-bottom, clear microtiter plates, sealed with a Breathe-Easy membrane (Merck KGaA), and incubated as described above. Specific growth rates were calculated using linear regression from the blank-subtracted exponential growth phase. Expression of IPTG-inducible GFP and T7 RNAP was performed in 96-well plates with a clear lid with a total volume of 200 µl. For this analysis, the strains were co-transformed with both expression plasmids and plated on LB with 1% glucose (to repress leaky expression) and for the unmodified strain HS996, antibiotics (Cm, Zeo). To create precultures, single colonies were inoculated into 200 µl LB (for HS996 both with and without antibiotics) in 96-well plates, sealed with a Breathe-Easy membrane and incubated overnight at 30 °C with shaking. The overnight cultures were then diluted 1:100 in a clear 96-well plate with fresh LB media for expression. The cultures were incubated at 30 °C with non-continuous orbital shaking (5 s on, 10 s off) for 7 h, induced with 1 mM IPTG and then incubated at 30 °C for a further 24 h. OD 600 was monitored with measurements every 20 min. Fluorescence measurements were performed every 20 min using the Fluorescein family filter pair (excitation 485 nm, and emission 525 nm), with filter bandwidths of 15 nm and 15 nm full width at half maximum (FWHM) for excitation and emission respectively.

Plasmid sequences and images

Plasmid maps and annotated DNA sequences were generated using SnapGene ® software (Dotmatics).

Rha-inducible infA expression for inducible complementation

Genome engineering for Rha-inducible infA expression

We first investigated whether P Rha can be used to enable inducible complementation of the essential gene infA. To create a Rha-dependent strain, we exchanged the endogenous infA promoter with a cassette containing P Rha and the transcriptional activators rhaS and rhaR using λ-Red recombineering, as described previously (Brechun et al., 2024). The final configuration is shown in Figure 1. All modifications were confirmed via PCR and sequencing. The resulting strain is termed HS996- P Rha - infA (SGB311).

Rha-dependent growth and plasmid-based infA complementation

First, we characterized the strain by monitoring inducer-dependent growth. The strain HS996- P Rha - infA (SGB311) was grown overnight in LB-Rha, diluted to an OD 600 of 0.02 and streaked on LB-agar with and without inducer supplementation. Growth was only observed on LB-Rha (Figure 2A). Likewise, growth in liquid media was examined. For this, an overnight culture grown in LB-Rha was diluted to an OD of 0.005 in LB, and then inoculated with a 1:10 dilution in LB and LB-Rha. As expected, robust growth was observed upon induction of infA expression (Figure 2B). Minor background growth was detected in the absence of Rha supplementation. This could be due to remaining pools of the translated infA gene product, or due to minor leaky expression from P Rha. However, since this background growth is limited and insignificant in comparison to the growth of the culture with an actively transcribed infA copy, we did not investigate this further.

Next, the behaviour of the modified strain with complementing plasmids was examined. The modified strain HS996- P Rha - infA (SGB311) was transformed with the model plasmid CGB237, which encodes a constitutively expressed infA cassette for complementation. A plasmid map is provided in the SI (Supplementary Figure S1). The transformation was performed antibiotic-free; selection was achieved by plating the cells on LB plates without Rha supplementation. Ten colonies were chosen at random from the transformation plate, and the plasmid content was analyzed using a restriction enzyme digestion. All colonies contained plasmid with the correct banding pattern (Figure 2C). Next, strain fitness was examined by monitoring the specific growth rate of HS996- P Rha - infA (SGB311) with CGB237 in comparison to the unmodified parental strain with the same plasmid. To ensure plasmid maintenance in both cultures, the unmodified strain was grown with antibiotic selection (Cm), while the modified strain was grown in non-supplemented media. No statistically significant difference in specific growth rate was detected (unpaired, two-sided t -test, α = 0.05), indicating that redirecting infA transcription to the complementing plasmid does not yield any growth defects (Figure 2D). To visually examine the selection pressure for plasmid uptake during transformation, a transformation was performed using a complementing plasmid encoding a constitutively expressed chromoprotein. The chromoprotein is composed of the genes for two fluorescent proteins (E2-Crimson and mStrawberry) genetically linked via a short synthetic linker. The expression of these two fluorescent proteins results in a purple pigment that is visible by eye after prolonged incubation (i.e. 48 h). A map of the model plasmid, termed CGB285, is shown in the SI (Supplementary Figure S2). The transformation was performed via complementation by plating the culture on LB media without Rha. After extended incubation of the plate, all colonies developed a purple pigment; the lack of pigment-free colonies indicates that only cells that had taken up the complementing plasmid were able to proliferate enough to form colonies (Figure 2E).

Plasmid maintenance in the Rha-inducible infA complementation strain

Next, antibiotic-free plasmid maintenance was examined using the model plasmid CGB327 (Supplementary Figure S3). This model plasmid has a high copy origin of replication (ColE1) and encodes an infA complementation cassette, a constitutively expressed fluorescent marker (mStrawberry), and a Cm resistance marker. The constitutively expressed fluorescent protein creates a metabolic load to mimic the expression of a model gene of interest. This is important for evaluation of plasmid stability, as it has been shown previously that uninduced plasmids are often stably maintained even in the absence of selection pressure (Khananisho et al., 2023). The metabolic load associated with plasmid-based protein expression has been shown to cause plasmid instability (Cumming et al., 2022; Khananisho et al., 2023), suggesting high productivity is intrinsically linked with a high propensity for plasmid loss. The model plasmid was transformed into HS996- P Rha - infA (SGB311) and into the parental strain HS996, using LB and LB-Cm respectively for plasmid selection. Next, single colonies were grown in LB media for 72 h, and plasmid maintenance was examined by plating culture dilutions and patching isolated colonies on different media, as described in detail the Materials and Methods. The analysis was performed three times for each strain.

Plating culture dilutions on different media allowed a direct examination of plasmid presence in the cultures after 72 h of growth (Figure 3A). The unmodified strain showed comparable growth on LB and LB-Rha plates, and reduced growth on LB-Cm, indicating a significant proportion of cells in the culture had lost the model plasmid carrying the Cm R cassette. The lack of pigment shows a loss in expression the plasmid-encoded chromogenic marker in non-selective conditions. The engineered strain showed comparable growth on all three plates and uniform pigmentation, indicating cells in the culture maintained the model plasmid throughout the 72 h growth trial. Plating culture dilutions on LB-Rha investigated if the modified strain cultures contained growth-arrested plasmid-free cells unable to propagate on LB; no increased growth was observed on the non-selective LB-Rha media.

Plasmid maintenance was also analyzed in single cells. The 72 h cultures were streaked on LB media to isolate single colonies (thus extending the total LB growth period to 96 h). Isolated colonies were selected at random (prior to pigment development) and patched on media with and without antibiotics. All colonies isolated from the HS996- P Rha - infA (SGB311) culture were found to contain plasmid, while substantial plasmid loss was observed in the wild-type strain (Figures 3B, C). Overall, these results demonstrate reliable antibiotic-free plasmid maintenance via infA complementation in the engineered strain HS996- P Rha - infA (SGB311) in non-supplemented media.

Inducible secY complementation

Genome engineering for inducible secY complementation

Next, a strain was developed to enable inducible complementation using the essential gene secY. The gene secY is a challenging target for modification because it is present in a locus with a high density of essential genes. In E. coli K-12, secY is encoded in an operon with 10 upstream essential genes as well as one downstream non-essential gene, and is immediately adjacent to an operon with 5 other essential genes. We therefore chose to insert an ectopic inducible secY copy, and to create a secY knock-out at the endogenous locus. For this, a synthetic fragment encoding secY under the control of P Ara, a Cm R cassette, and the repressor araC were integrated into the recAX locus (Figure 4A). Next, the endogenous secY was deleted through insertion of a flippase recognition target (FRT)-flanked Zeo R cassette; at this point, the strain was supplemented with Ara to induce expression of the ectopic secY (Figure 4B). Following these modifications, all antibiotic markers were removed to create a strain without ARGs. The Zeo R marker in the endogenous secY locus was removed using Flp-recombinase. The Cm R in the recAX locus was removed by replacing it with a loxP-flanked Zeo R gene, which was subsequently eliminated using Cre-recombinase. This strain is referred to as HS996- P Ara - secY (SGB307). The insertion of the Ara-inducible secY and the deletion of the endogenous secY were confirmed with PCR (Figure 4C) and sequencing.

Ara-dependent growth and plasmid-based secY complementation

The Ara-dependent growth of the modified strain HS996- P Ara - secY (SGB307) was confirmed by testing growth on solid media and in liquid culture with and without supplementation of the inducer, Ara (Figures 5A, B). Robust growth was observed only with Ara supplementation, confirming the Ara-dependent growth phenotype.

Next, the behaviour of the modified strain HS996- P Ara - secY (SGB307) with complementing plasmids was examined. First, a plasmid was assembled containing secY under a constitutive promoter, a high copy origin of replication (ColE1) and a Zeo R gene. This plasmid is referred to as CGB326 (Supplementary Figure S4). The plasmid was transformed antibiotic-free into HS996- P Ara - secY (SGB307); plasmid selection was achieved by plating the cells on LB without Ara. Ten colonies from the transformation were picked at random, and the plasmid content was analyzed using a restriction enzyme digestion. All colonies contained plasmid with the correct banding pattern (Figure 5C). Next, strain fitness was examined by comparing the specific growth rate of the modified strain HS996- P Ara - secY (SGB307) with the unmodified strain HS996, each transformed with CGB326. The unmodified strain HS996 was grown using antibiotic selection pressure (Zeo) to ensure plasmid maintenance. No significant difference in the growth rate of the modified and unmodified strains was detected (unpaired, two-sided t -test, α = 0.05) (Figure 5D). Next, the selection pressure for plasmid uptake during transformation was examined visually using a secY -complementing plasmid encoding a chromogenic marker. For this, the plasmid CGB332 was developed (Supplementary Figure S5), which contains a constitutively expressed chromoprotein on a high-copy plasmid backbone (ColE1), a Zeo R marker and a secY complementation marker. This model plasmid was transformed into HS996- P Ara - secY (SGB307) and plated on non-supplemented LB media. After extended incubation, all colonies produced purple pigmentation, demonstrating that only cells that had taken up plasmid could proliferate enough to form colonies (Figure 5E).

Plasmid maintenance in the Ara-inducible secY complementation strain

Next, antibiotic-free plasmid maintenance via secY complementation was monitored using the model plasmid CGB332. CGB332 was transformed into HS996- P Ara -secY (SGB307) and into the unmodified strain HS996 using LB and LB-Zeo respectively for plasmid selection. Single colonies of each strain were grown in LB media without supplementation for 72 h, and plasmid maintenance was analyzed as described in section 3.1.3., with the exception that LB-Ara and LB-Zeo plates were used (instead of LB-Rha and LB-Cm plates respectively). Overall, the results revealed significant plasmid loss in the unmodified cultures, and stable plasmid maintenance in the engineered strain (Figure 6). The dilution series of the engineered strain (‘ secY ’) revealed comparable growth and pigmentation on LB, LB-Ara and LB-Zeo plates, indicating the cells in the culture maintained the plasmid during the 72 h growth trial. No increased growth was observed on LB-Ara dilutions (non-selective conditions for the modified strain). In contrast, the unmodified strain (‘wt’) showed significantly less growth on LB-Zeo and a loss of pigmentation in non-selective conditions, indicating significant plasmid loss. Likewise, the single cell analysis showed stable plasmid maintenance in the engineered strain and dramatic plasmid loss in the unmodified strain. All colonies examined from HS996- P Ara -secY (SGB307) contained plasmid, and all colonies examined from HS996 had lost the plasmid. As before, isolated colonies were patched on plates with and without antibiotic prior to development of pigment in the colonies.

Simultaneous antibiotic-free plasmid maintenance of two plasmids

Strain design and characterization

The promising results from the secY complementation allowed us to combine inducible secY and inducible infA complementation for simultaneous antibiotic-free maintenance of two plasmids, one expressing infA and a second expressing secY. For this, we generated two strains: one strain with infA and secY each under the control of P Ara, and a second strain with infA under the control of P Rha and secY under the control of P Ara. To generate the first strain, we further modified the strain HS996- P Ara - secY (SGB307) by exchanging the endogenous infA promoter with P Ara as we described previously (Brechun et al., 2024). This created a strain that requires Ara supplementation, or transformation with both an infA and a secY complementing plasmid. To generate the second strain, we further modified the strain HS996- P Ara - secY (SGB307) by exchanging the endogenous infA promoter with P Rha as we described above. In this strain, expression of each essential gene infA and secY can be independently controlled; therefore, in addition to growth with and without both plasmids, this design should allow the strain to be propagated with a single plasmid by supplementing with the corresponding inducer in the absence of the complementing plasmid (i.e. in the absence of an infA complementing plasmid, the strain could be supplemented with Rha to induce genomic infA expression). As before, we removed all antibiotic resistance markers used to generate the strains, resulting in ARG-free modified strains. These strains are referred to as HS996- P Ara -infA-P Ara -secY (SGB308) and HS996- P Rha -infA-P Ara -secY (SGB312) respectively.

Inducer dependent growth and complementation with multiple plasmids

First, we characterized the strains by monitoring inducer-dependent growth. The strains HS996- P Ara -infA-P Ara -secY (SGB308) and HS996- P Rha -infA-P Ara -secY (SGB312) were grown overnight in LB with the appropriate inducer(s) (Ara and Ara-Rha respectively); the following morning, the strains were diluted to an OD 600 of 0.005 in LB and inoculated with a 1:10 dilution in LB and LB with inducer(s). Growth of each strain was only observed with inducer supplementation (Figures 7A, B). Likewise, inducer-dependent growth was monitored on solid media by diluting an overnight culture to an OD 600 of 0.02, then streaking the cultures on LB-agar with and without inducer supplementation. This confirmed inducer-dependent growth in the absence of a complementing plasmid on solid media (Figures 7C, D column ‘none’).

The two orthogonal inducible promoters in the strain HS996- P Rha -infA-P Ara -secY (SGB312) allow the two essential genes to be independently induced; this therefore enables the strain to be propagated with a single complementing plasmid, by supplementing the culture with one inducer. To demonstrate this, the two compatible model plasmids CGB327 (Supplementary Figure S3) and CGB331 (Supplementary Figure S6) were used. These plasmids each encode an essential gene (infA or secY), a constitutively expressed model gene of interest (mStrawberry and mEmerald) and an antibiotic resistance marker (Cm R or Zeo R). Both plasmids were co-transformed into HS996- P Rha -infA-P Ara -secY (SGB312), using non-supplemented LB for plasmid selection. In addition, the strain was transformed with each plasmid individually; here, the cultures were plated on media where one inducer was omitted to select for the complementing plasmid and the second inducer was supplemented, to activate genomic expression in the absence of a complementing plasmid. Specifically, to transform HS996- P Rha -infA-P Ara -secY (SGB312) with CGB327 (infA complementing plasmid), cultures were plated on LB-Ara (inducing genomic secY expression), and to transform the strain with CGB331 (secY complementing plasmid), cultures were plated on LB-Rha (inducing genomic infA expression). From each transformation, a single colony was picked and propagated in LB with inducer as necessary (i.e. replicating the transformation conditions). Next, the overnight cultures were diluted to an OD 600 of 0.02 and streaked on LB-agar with and without inducer supplementation. The strain showed growth only when conditions allowed for expression of both infA and secY, either through plasmid complementation or inducer supplementation (Figure 7D).

Next, we compared the specific growth rate of each modified strain, HS996- P Ara -infA-P Ara - secY (SGB308) and HS996- P Rha -infA-P Ara -secY (SGB312), and the unmodified parental strain. For this analysis, we transformed each strain with two compatible complementing plasmids: CGB237 (infA on a ColE1 plasmid; Supplementary Figure S1) and CGB334 (secY on a p15A plasmid; Supplementary Figure S7). The plasmids were co-transformed into the two modified strains using a lack of inducer supplementation for plasmid selection, and into the unmodified parental strain using antibiotic selection (Cm, Zeo). Growth rates were monitored in biological triplicate (Figure 8). No statistically significant difference was detected between any of the growth rates (one-way ANOVA, α = 0.05), indicating redirecting transcription of the essential genes from the genome to the complementing plasmids does not result in a significant physiological change.

Stable, antibiotic-free maintenance of two plasmids

Next, we examined plasmid maintenance using two compatible complementing plasmids, each with a constitutively expressed fluorescent marker as a model gene of interest. Plasmid maps are shown in the SI (CGB327 Supplementary Figure S3 and CGB328 Supplementary Figure S8), and a schematic is shown in Figure 9A. We co-transformed the engineered strains HS996- P Ara -infA-P Ara -secY (SGB308) and HS996- P Rha -infA-P Ara -secY (SGB312) using a lack of inducer for plasmid selection, and the unmodified strain HS996 using antibiotic selection (Zeo, Cm). Plasmid maintenance was examined in single cells as described previously, with the exception that the liquid culture growth period was shortened to two days, as it was anticipated that the two plasmids would cause a higher metabolic burden and result in more rapid plasmid loss. After 48 h of growth in non-supplemented media, the cultures were streaked on LB-agar to obtain single colonies. Isolated colonies were patched on LB, LB-Cm and LB-Zeo prior to pigment development to test for the presence of each plasmid (Figures 9B, C). The unmodified strain showed unreliable plasmid maintenance after growth in non-supplemented media. In particular, the high copy plasmid CGB327 was lost rapidly, as shown by a loss of Cm resistance. This is consistent with the expectation that constitutive expression of a fluorescent protein from a high copy plasmid would exert a high metabolic load. Overall, only approximately one third of colonies from the unmodified strain contained both plasmids after the growth trial. In contrast, all colonies from both modified strains showed growth on Zeo and Cm, indicating both plasmids were stably maintained in non-supplemented LB media.

Protein expression using two antibiotic-free plasmids

Next, antibiotic-free protein expression from a two-plasmid system was evaluated. For this analysis, we used the plasmids CGB323 (encoding mEmerald expression from a T7 promoter, Cm R and infA), and CGB342 (encoding T7-RNA polymerase expression from the lac promoter, Zeo R and secY). Plasmid maps are shown in the SI (Supplementary Figures S9, S10). This creates a simple test system where both plasmids must be present for expression of the fluorescent marker to occur. We chose to use a T7-based expression system, because expression from a T7 promoter is known to lead to plasmid instability due to the metabolic burden from the high expression level (Khananisho et al., 2023). We co-transformed the plasmids into the engineered strains using infA and secY complementation, and into the unmodified strain using antibiotic-based selection. The engineered strains were expressed in antibiotic-free LB media, while the unmodified strain was expressed both in antibiotic-free LB media and in LB media with antibiotic supplementation (Figure 10). This allows the expressions using conditional complementation to be compared with an expression under non-selective conditions and an expression performed under the state-of-the-art selection conditions.

After 5 hours of expression, there was no significant difference in the productivity of the three cultures grown with plasmid selection conditions (i.e. complementation for the engineered strains in non-supplemented media, and antibiotic-selection pressure for the unmodified strain). All cultures with a form of plasmid selection pressure were significantly more productive than HS996 in antibiotic-free media (Figure 10B; one-way ANOVA followed by a Tukey’s HSD post-hoc test).

After 24 h of expression, the two modified strains showed comparable expression levels, and significantly more fluorescence than the unmodified strain both in media with and without antibiotics (Figure 10C; one-way ANOVA followed by a Tukey’s HSD post-hoc test). This indicates a decrease in the productivity of the unmodified strain despite antibiotic selection. This likely is a result of a decrease in the stringency of the antibiotic selection pressure due to enzymatic inactivation by the plasmid-encoded ARGs. Indeed, the HS996 cultures in antibiotic-supplemented media shows a significant drop in fluorescence over time, consistent with the proliferation of non-productive cells that have lost one or both plasmids.

These results show that inducible complementation provides plasmid maintenance pressure that is equally as effective as conventional antibiotic-based selection for short-term expressions and is potentially superior for long-term expressions. In contrast to antibiotic-based selection pressure, which may decrease over time due to antibiotic inactivation, the selection pressure imposed by complementation remains constant throughout the expression. The selection pressure imposed by conditional complementation is highly robust because the design selects for cells expressing a beneficial gene, which is difficult to bypass. For instance, the selection pressure cannot be bypassed by simple inactivating mutations, as is the case in toxin-antitoxin systems, where any mutation inactivating toxin expression defuses the selection pressure. Here, in order for the modified strain to bypass the selection pressure (and thus grow in the absence of a complementing plasmid or the inducer), the plasmid would need to integrate into the genome, or a mutation would need to occur to render the genomic inducible promoter (P Ara or P Rha) constitutively active. We have not observed genome integration of the complementing plasmids. The small size of the infA gene makes such a recombination event unlikely. However, if such recombination proves to be problematic, this could be eliminated by using alternative codons in one of the essential gene copies to reduce homology between the genomic and the plasmid copies. Specific, function-tuning mutations able to render an inducible promoter constitutively active are not likely to arise with a high frequency. Overall, the plasmid selection pressure is expected to have high stability.

It is also worth noting that this comparison was performed using the same plasmids in the modified and unmodified strains; however, the engineered strains do not require the ARGs, and the unmodified strain does not require the complementing genes. In the case of the engineered strains, the plasmid-encoded ARGs represent unnecessary additional heterologous expression cassettes. In contrast, the complementation markers would be expressed by the strain regardless of plasmid use; the transcription is simply being redirected from the genome to the plasmid for use as a marker. It has been shown that ARGs are often expressed at an unnecessarily high level (Cumming et al., 2022), which exerts a metabolic burden that can decrease the product yield (Rozkov et al., 2004; Cunningham et al., 2009). Therefore, in the absence of the ARGs, the complementation strains should have increased metabolic capacity that could be channeled to produce the protein of interest. Since our data also show increased selection pressure stability, this suggests that plasmid maintenance via complementation could overall be better suited than conventional antibiotic-based selection for protein production.

Antibiotic-free biomanufacturing is becoming increasingly attractive due to regulatory pressure and expenses associated with industrial-scale antibiotic use. Our results demonstrate that conditional complementation can be used for the stable maintenance of multiple plasmids in non-supplemented media without any loss in performance. We show that two inducible promoters (P Ara and P Rha) and two essential genes (infA and secY) can be used for efficient plasmid maintenance, and that these modifications can be combined to enable ARG-free maintenance of multiple plasmids. This highlights the flexibility of this strategy. We expect that conditional complementation for plasmid maintenance could be achieved using other combinations of promoters (natural or engineered) and essential genes. Furthermore, it should be possible to extend this design to enable antibiotic-free selection of more than two plasmids by using additional essential genes.

We previously demonstrated conditional complementation using a P Ara -infA design in a wide range of E. coli strains (Brechun et al., 2026), including common K12 and B strains (such as MG1655 and BL21), and more phylogenetically diverse strains (Nissle1917 and W). The design was fully functional in all strain backgrounds. Notably, the system was effective regardless of whether the strains could metabolize the inducer (Ara), showing the remarkably robust nature of the design. We therefore anticipate that conditional complementation using the alternative promoter (P Rha), the alternative essential gene (secY) and the combination thereof should be transferrable to other E. coli strains for single or dual plasmid maintenance. Likewise, we expect the design could be implemented in other important production microorganisms, such as Bacillus, Corynebacterium or Pseudomonas, if a sufficiently tight inducible promoter is available to control the expression of a non-metabolic essential gene.

The essential component of this strategy is the ability to rescue the plasmid-free engineered strain using inducer supplementation. This provides two important advantages: the engineered strains can be transformed with new plasmids with simplicity equal to conventional antibiotic-based transformations, and the transformations can be performed antibiotic-free, allowing plasmids to be used that are devoid of an ARG. The first point offers significantly improved handling. We demonstrate that by using two orthogonal inducible promoters, it is even possible to create a strain where one or two complementing plasmids can be used. The second point reduces the metabolic load of the plasmid and improves the sustainability of biomanufacturing. Eliminating plasmid-encoded ARGs eliminates the risk of contributing to AMR through horizontal gene transfer (HGT). A recent study has shown that standard methods for decontaminating biological waste do not fully inactivate DNA (Gluth et al., 2026), highlighting the risk that plasmid-encoded ARGs from synthetic strains could spread via HGT. The current study provides a robust and flexible solution for plasmid maintenance of multiple plasmids that eliminates the need for antibiotics and ARGs without imposing any technical restrictions or compromising handling.